diff --git a/enterprise/litellm_enterprise/proxy/liteadmin.py b/enterprise/litellm_enterprise/proxy/liteadmin.py
index 6a9110f1460..e89bf11bd6e 100644
--- a/enterprise/litellm_enterprise/proxy/liteadmin.py
+++ b/enterprise/litellm_enterprise/proxy/liteadmin.py
@@ -15,11 +15,12 @@ from urllib.parse import urlencode, urlsplit
 import httpx
 from fastapi import APIRouter, Depends, HTTPException, Request
 from fastapi.responses import HTMLResponse, RedirectResponse, Response
-from pydantic import BaseModel, ConfigDict, Field, SecretStr, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, SecretStr, TypeAdapter, ValidationError
 
 from litellm.llms.custom_httpx.http_handler import get_async_httpx_client
 from litellm.proxy._experimental.mcp_server.oauth_utils import get_request_base_url
 from litellm.proxy._types import LiteLLM_UserTable, LitellmUserRoles, UserAPIKeyAuth
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.auth.auth_checks import UserNotFoundError
 
 router: Final = APIRouter()
@@ -34,14 +35,14 @@ _HEADERS: Final = {
 }
 
 
-class LinkDetails(BaseModel):
+class LinkDetails(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, strict=True, extra="forbid")
     workspace_id: str = Field(min_length=1, max_length=64)
     slack_user_id: str = Field(min_length=1, max_length=64)
     email: str = Field(min_length=1, max_length=320)
 
 
-class AdminSession(BaseModel):
+class AdminSession(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     user_id: str
     credential: SecretStr
diff --git a/enterprise/litellm_enterprise/types/enterprise_callbacks/send_emails.py b/enterprise/litellm_enterprise/types/enterprise_callbacks/send_emails.py
index 380b0a6facb..a2322d7f650 100644
--- a/enterprise/litellm_enterprise/types/enterprise_callbacks/send_emails.py
+++ b/enterprise/litellm_enterprise/types/enterprise_callbacks/send_emails.py
@@ -1,12 +1,13 @@
 import enum
 from typing import Dict, List, Optional
 
-from pydantic import BaseModel, Field
+from pydantic import Field
 
 from litellm.proxy._types import WebhookEvent
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class EmailParams(BaseModel):
+class EmailParams(LiteLLMBaseModel):
     logo_url: str
     support_contact: str
     base_url: str
@@ -39,14 +40,14 @@ class EmailEvent(str, enum.Enum):
     soft_budget_crossed = "Soft Budget Crossed"
     max_budget_alert = "Max Budget Alert"
 
-class EmailEventSettings(BaseModel):
+class EmailEventSettings(LiteLLMBaseModel):
     event: EmailEvent
     enabled: bool
-class EmailEventSettingsUpdateRequest(BaseModel):
+class EmailEventSettingsUpdateRequest(LiteLLMBaseModel):
     settings: List[EmailEventSettings]
-class EmailEventSettingsResponse(BaseModel):
+class EmailEventSettingsResponse(LiteLLMBaseModel):
     settings: List[EmailEventSettings]
-class DefaultEmailSettings(BaseModel):
+class DefaultEmailSettings(LiteLLMBaseModel):
     """Default settings for email events"""
     settings: Dict[EmailEvent, bool] = Field(
         default_factory=lambda: {
diff --git a/enterprise/litellm_enterprise/types/proxy/audit_logging_endpoints.py b/enterprise/litellm_enterprise/types/proxy/audit_logging_endpoints.py
index 4615bde2b15..3676cf2a7a2 100644
--- a/enterprise/litellm_enterprise/types/proxy/audit_logging_endpoints.py
+++ b/enterprise/litellm_enterprise/types/proxy/audit_logging_endpoints.py
@@ -1,10 +1,12 @@
 from datetime import datetime
 from typing import Any, Dict, List, Optional
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class AuditLogResponse(BaseModel):
+class AuditLogResponse(LiteLLMBaseModel):
     """Response model for a single audit log entry"""
 
     id: str
@@ -18,7 +20,7 @@ class AuditLogResponse(BaseModel):
     updated_values: Optional[Dict[str, Any]] = None
 
 
-class PaginatedAuditLogResponse(BaseModel):
+class PaginatedAuditLogResponse(LiteLLMBaseModel):
     """Response model for paginated audit logs"""
 
     audit_logs: List[AuditLogResponse]
diff --git a/litellm/caching/caching_handler.py b/litellm/caching/caching_handler.py
index f04ff8b6e78..9d02c84f6e8 100644
--- a/litellm/caching/caching_handler.py
+++ b/litellm/caching/caching_handler.py
@@ -21,7 +21,7 @@ import time
 from collections.abc import AsyncGenerator, AsyncIterator, Awaitable, Callable, Generator, Mapping
 from typing import TYPE_CHECKING, Any, Final, Optional, TypeVar
 
-from pydantic import BaseModel, ConfigDict, ValidationError
+from pydantic import ConfigDict, ValidationError
 
 import litellm
 from litellm._internal_context import post_response_phase
@@ -37,6 +37,7 @@ from litellm.litellm_core_utils.logging_utils import (
 )
 from litellm.types.caching import EMBEDDING_CACHE_FORMAT_VERSION, CachedEmbedding
 from litellm.types.integrations.custom_logger import converted_stream_requested
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import ResponsesAPIResponse
 from litellm.types.rerank import RerankResponse
 from litellm.types.utils import (
@@ -68,7 +69,7 @@ from litellm.litellm_core_utils.core_helpers import (
 from litellm.litellm_core_utils.streaming_handler import CustomStreamWrapper
 
 
-class CachingHandlerResponse(BaseModel):
+class CachingHandlerResponse(LiteLLMBaseModel):
     """
     This is the response object for the caching handler. We need to separate embedding cached responses and (completion / text_completion / transcription) cached responses
 
@@ -172,7 +173,7 @@ def _request_cache_key(request_kwargs: Mapping[str, Any]) -> str | None:
     return request_kwargs.get("cache_key", None)
 
 
-class _CachedEmbeddingRecord(BaseModel):
+class _CachedEmbeddingRecord(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     embedding: list[float] | str | None
diff --git a/litellm/constants.py b/litellm/constants.py
index a268f00775a..938a3c85c29 100644
--- a/litellm/constants.py
+++ b/litellm/constants.py
@@ -5,6 +5,7 @@ from typing import Final, Literal
 
 from litellm.litellm_core_utils.env_utils import get_env_int, get_env_int_in_range, get_env_int_or_none
 
+DEFER_PYDANTIC_BUILD: Final = os.getenv("DEFER_PYDANTIC_BUILD", "true") in ("true", "1", "on")
 DEFAULT_HEALTH_CHECK_PROMPT: Final = str(os.getenv("DEFAULT_HEALTH_CHECK_PROMPT", "test from litellm"))
 AZURE_DEFAULT_RESPONSES_API_VERSION: Final = str(os.getenv("AZURE_DEFAULT_RESPONSES_API_VERSION", "preview"))
 AZURE_OPENAI_AUDIO_PROVIDERS: Final = frozenset({"azure", "azure_ai"})
diff --git a/litellm/cost_calculator.py b/litellm/cost_calculator.py
index 0443faabe59..8971b958c1c 100644
--- a/litellm/cost_calculator.py
+++ b/litellm/cost_calculator.py
@@ -100,7 +100,7 @@ from litellm.llms.xai.cost_calculator import cost_per_token as xai_cost_per_toke
 from litellm.responses.utils import ResponseAPILoggingUtils
 from litellm.types.agents import LiteLLMSendMessageResponse
 from litellm.types.decisions import DecisionsResponse, DecisionsUsage
-from litellm.types.llms.base import CachedTokensDetails
+from litellm.types.llms.base import CachedTokensDetails, LiteLLMBaseModel
 from litellm.types.llms.openai import (
     HttpxBinaryResponseContent,
     ImageGenerationRequestQuality,
@@ -2839,12 +2839,12 @@ class RealtimeAPITokenUsageProcessor(BaseTokenUsageProcessor):
 _RESPONSES_WS_BILLABLE_EVENT_TYPES: Final = frozenset({"response.completed", "response.incomplete"})
 
 
-class _ResponsesWsEventResponse(BaseModel):
+class _ResponsesWsEventResponse(LiteLLMBaseModel):
     usage: Mapping[str, object] | None = None
     service_tier: str | None = None
 
 
-class _ResponsesWsEvent(BaseModel):
+class _ResponsesWsEvent(LiteLLMBaseModel):
     type: str = ""
     response: _ResponsesWsEventResponse | None = None
 
diff --git a/litellm/google_genai/main.py b/litellm/google_genai/main.py
index c1822e4720d..9ab3d78471c 100644
--- a/litellm/google_genai/main.py
+++ b/litellm/google_genai/main.py
@@ -5,7 +5,7 @@ from functools import partial
 from typing import TYPE_CHECKING, Any, ClassVar, Final
 
 import httpx
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 
 import litellm
 from litellm.constants import request_timeout
@@ -17,6 +17,7 @@ from litellm.llms.base_llm.google_genai.transformation import (
     BaseGoogleGenAIGenerateContentConfig,
 )
 from litellm.llms.custom_httpx.llm_http_handler import BaseLLMHTTPHandler
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.router import GenericLiteLLMParams
 from litellm.types.utils import CallTypes
 from litellm.utils import ProviderConfigManager, client
@@ -46,7 +47,7 @@ def _mark_async_entrypoint(logging_obj: LiteLLMLoggingObj | None, marker: str, i
         logging_obj.model_call_details.setdefault("litellm_params", {})[marker] = is_async
 
 
-class GenerateContentSetupResult(BaseModel):
+class GenerateContentSetupResult(LiteLLMBaseModel):
     """Internal Type - Result of setting up a generate content call"""
 
     model_config: ClassVar[ConfigDict] = ConfigDict(arbitrary_types_allowed=True)
diff --git a/litellm/harness/handlers/tool_loop_handler.py b/litellm/harness/handlers/tool_loop_handler.py
index d3acb50eadb..e3276793fcb 100644
--- a/litellm/harness/handlers/tool_loop_handler.py
+++ b/litellm/harness/handlers/tool_loop_handler.py
@@ -12,7 +12,7 @@ from dataclasses import dataclass
 from types import MappingProxyType
 from typing import Final, Protocol, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, TypeAdapter, ValidationError
+from pydantic import ConfigDict, TypeAdapter, ValidationError
 
 import litellm
 from litellm.harness.context import SessionContext
@@ -28,6 +28,7 @@ from litellm.llms.tool_loop.harness.transformation import (
     function_tool,
 )
 from litellm.types.completion import ChatCompletionMessageParam
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import (
     ChatCompletionMessageCustomToolCall,
     ChatCompletionMessageToolCall,
@@ -52,7 +53,7 @@ _JSON_DECODER: Final = json.JSONDecoder()
 _HISTORY_ADAPTER: Final = TypeAdapter(list[dict[str, object]])
 
 
-class _Usage(BaseModel):
+class _Usage(LiteLLMBaseModel):
     model_config = ConfigDict(from_attributes=True)
 
     prompt_tokens: int | None = None
diff --git a/litellm/integrations/deepeval/types.py b/litellm/integrations/deepeval/types.py
index c86d01d0468..eb4bb1f9b2a 100644
--- a/litellm/integrations/deepeval/types.py
+++ b/litellm/integrations/deepeval/types.py
@@ -2,7 +2,9 @@
 from enum import Enum
 from typing import Any, ClassVar, Literal
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
 class SpanApiType(Enum):
@@ -21,7 +23,7 @@ class TraceSpanApiStatus(Enum):
     ERRORED = "ERRORED"
 
 
-class BaseApiSpan(BaseModel):
+class BaseApiSpan(LiteLLMBaseModel):
     model_config: ClassVar[ConfigDict] = ConfigDict(use_enum_values=True)
 
     uuid: str
@@ -44,7 +46,7 @@ class BaseApiSpan(BaseModel):
     cost_per_output_token: float | None = Field(None, alias="costPerOutputToken")
 
 
-class TraceApi(BaseModel):
+class TraceApi(LiteLLMBaseModel):
     uuid: str
     base_spans: list[BaseApiSpan] = Field(alias="baseSpans")
     agent_spans: list[BaseApiSpan] = Field(alias="agentSpans")
diff --git a/litellm/integrations/galileo.py b/litellm/integrations/galileo.py
index 9bc324256fb..56147cd720d 100644
--- a/litellm/integrations/galileo.py
+++ b/litellm/integrations/galileo.py
@@ -9,7 +9,7 @@ from datetime import datetime, timezone, tzinfo
 from typing import Any, Final, Protocol, cast
 
 import httpx
-from pydantic import BaseModel, ConfigDict, Field, TypeAdapter
+from pydantic import ConfigDict, Field, TypeAdapter
 from typing_extensions import ReadOnly, TypedDict
 
 import litellm
@@ -24,6 +24,7 @@ from litellm.llms.custom_httpx.http_handler import (
     httpxSpecialProvider,
 )
 from litellm.types.integrations.base_health_check import IntegrationHealthCheckStatus
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import (
     AllMessageValues,
     HttpxBinaryResponseContent,
@@ -80,7 +81,7 @@ class GalileoStandardLoggingFields(TypedDict, total=False):
     endTime: float
 
 
-class LLMResponse(BaseModel):
+class LLMResponse(LiteLLMBaseModel):
     latency_ms: int
     status_code: int
     input_text: str
diff --git a/litellm/integrations/langfuse/langfuse_sdk.py b/litellm/integrations/langfuse/langfuse_sdk.py
index 986f35297d2..d117120dd70 100644
--- a/litellm/integrations/langfuse/langfuse_sdk.py
+++ b/litellm/integrations/langfuse/langfuse_sdk.py
@@ -34,13 +34,14 @@ from opentelemetry.sdk.trace.id_generator import RandomIdGenerator
 from opentelemetry.sdk.trace.sampling import ALWAYS_ON, Decision, Sampler, SamplingResult
 from opentelemetry.trace import Link, NonRecordingSpan, Span, SpanContext, SpanKind, TraceFlags, Tracer, TraceState
 from opentelemetry.util.types import Attributes, AttributeValue
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 
 import litellm
 from litellm._logging import verbose_logger
 from litellm.integrations.langfuse.langfuse import PROMPT_CACHE_TTL_ENV, parse_langfuse_debug, whole_number
 from litellm.litellm_core_utils.safe_json_dumps import safe_dumps
 from litellm.llms.custom_httpx.http_handler import HTTPHandler, _get_httpx_client
+from litellm.types.llms.base import LiteLLMBaseModel
 
 __all__ = (
     "AuthCheckFailure",
@@ -1063,7 +1064,7 @@ def _auth_check_failure(reason: str) -> AuthCheckFailure:
     return AuthCheckFailure(reason)
 
 
-class _ApiErrorDetail(BaseModel):
+class _ApiErrorDetail(LiteLLMBaseModel):
     """The status and body of an ``ApiError``, whose own ``str`` also dumps every response header."""
 
     model_config = ConfigDict(frozen=True, from_attributes=True)
diff --git a/litellm/integrations/otel/model/config.py b/litellm/integrations/otel/model/config.py
index 6e9038c0cad..3cf49ea2f11 100644
--- a/litellm/integrations/otel/model/config.py
+++ b/litellm/integrations/otel/model/config.py
@@ -4,7 +4,7 @@ from enum import Enum
 from functools import lru_cache
 from typing import Annotated, Final
 
-from pydantic import AliasChoices, BaseModel, Field, TypeAdapter, ValidationError, field_validator, model_validator
+from pydantic import AliasChoices, ConfigDict, Field, TypeAdapter, ValidationError, field_validator, model_validator
 from pydantic.fields import FieldInfo
 from pydantic_settings import BaseSettings, NoDecode, PydanticBaseSettingsSource, SettingsConfigDict
 
@@ -15,6 +15,7 @@ from litellm.integrations.otel.model.baggage import (
     DEFAULT_BAGGAGE_TEAM_METADATA_KEYS,
 )
 from litellm.integrations.otel.model.spans import POSTGRESQL, db_system
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import OtelSpanScope
 
 #: Master feature-flag env var. The logger is inert until this is truthy.
@@ -62,7 +63,7 @@ def is_otel_v2_enabled() -> bool:
     return _OTelV2Flag().enabled
 
 
-class ExporterSpec(BaseModel):
+class ExporterSpec(LiteLLMBaseModel):
     """One span-export destination.
 
     The shared ``TracerProvider`` attaches one ``SpanProcessor`` per spec, so
@@ -70,7 +71,7 @@ class ExporterSpec(BaseModel):
     Phoenix + your own Honeycomb).
     """
 
-    model_config = {"extra": "forbid"}
+    model_config = ConfigDict(extra="forbid")
 
     kind: str = Field(
         default="console",
diff --git a/litellm/integrations/otel/model/destination.py b/litellm/integrations/otel/model/destination.py
index c9c035f24a1..2b25ef16347 100644
--- a/litellm/integrations/otel/model/destination.py
+++ b/litellm/integrations/otel/model/destination.py
@@ -8,12 +8,13 @@ from collections.abc import Mapping
 from typing import Final
 from urllib.parse import quote
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
 
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import OtelSpanScope
 
 
-class OtelDestination(BaseModel):
+class OtelDestination(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     endpoint: str
diff --git a/litellm/integrations/otel/model/request_io.py b/litellm/integrations/otel/model/request_io.py
index a315dadba2d..4131f3de6da 100644
--- a/litellm/integrations/otel/model/request_io.py
+++ b/litellm/integrations/otel/model/request_io.py
@@ -1,12 +1,13 @@
 from collections.abc import Mapping, Sequence
 from typing import Final, Literal
 
-from pydantic import BaseModel, ConfigDict, Field, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, TypeAdapter, ValidationError
 from typing_extensions import ReadOnly, TypedDict
 
 import litellm
 from litellm.integrations.otel.mappers.utils import json_or_none
 from litellm.proxy.guardrails.anthropic_sse import assemble_anthropic_sse_stream, is_raw_sse_stream
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import ResponseCompletedEvent, ResponsesAPIResponse
 from litellm.types.utils import ModelResponse, ModelResponseStream
 
@@ -20,7 +21,7 @@ class _Turn(TypedDict):
     content: ReadOnly[object]
 
 
-class _AnthropicMessage(BaseModel):
+class _AnthropicMessage(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     type: Literal["message"] = Field(exclude=True)
diff --git a/litellm/integrations/pointfive/upload_client.py b/litellm/integrations/pointfive/upload_client.py
index d3708d48661..8c4a281d04b 100644
--- a/litellm/integrations/pointfive/upload_client.py
+++ b/litellm/integrations/pointfive/upload_client.py
@@ -24,6 +24,7 @@ from litellm.types.integrations.pointfive import (
     PointFiveUploadFailure,
     PointFiveUploadTarget,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 UPLOAD_KIND: Final = "LITELLM"
 UPLOAD_URL_PATH: Final = "/upload-url"
@@ -31,21 +32,21 @@ PING_PATH: Final = "/ping"
 PUT_HEADERS: Final = MappingProxyType({"Content-Type": "application/x-ndjson", "Content-Encoding": "gzip"})
 
 
-class _PresignRequest(BaseModel):
+class _PresignRequest(LiteLLMBaseModel):
     kind: str = UPLOAD_KIND
     byte_count: int = Field(serialization_alias="byteCount")
 
 
-class _PingRequest(BaseModel):
+class _PingRequest(LiteLLMBaseModel):
     kind: str = UPLOAD_KIND
 
 
-class _TargetPayload(BaseModel):
+class _TargetPayload(LiteLLMBaseModel):
     upload_url: str = Field(alias="uploadUrl")
     object_key: str = Field(alias="objectKey")
 
 
-class _ErrorPayload(BaseModel):
+class _ErrorPayload(LiteLLMBaseModel):
     error: str = ""
 
 
diff --git a/litellm/integrations/prometheus_helpers/prometheus_api.py b/litellm/integrations/prometheus_helpers/prometheus_api.py
index e111474bd4d..f8cdb2353b6 100644
--- a/litellm/integrations/prometheus_helpers/prometheus_api.py
+++ b/litellm/integrations/prometheus_helpers/prometheus_api.py
@@ -7,7 +7,7 @@ import time
 from datetime import datetime, timedelta
 from typing import Final
 
-from pydantic import BaseModel, TypeAdapter
+from pydantic import TypeAdapter
 from typing_extensions import ReadOnly, TypedDict
 
 from litellm import get_secret
@@ -16,6 +16,7 @@ from litellm.llms.custom_httpx.http_handler import (
     get_async_httpx_client,
     httpxSpecialProvider,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 PROMETHEUS_URL: Final[str | None] = get_secret("PROMETHEUS_URL")
 PROMETHEUS_SELECTED_INSTANCE: Final[str | None] = get_secret("PROMETHEUS_SELECTED_INSTANCE")
@@ -24,18 +25,18 @@ async_http_handler: Final = get_async_httpx_client(llm_provider=httpxSpecialProv
 _RAW_JSON_PAYLOAD: Final = TypeAdapter(object)
 
 
-class PrometheusRangeSample(BaseModel):
+class PrometheusRangeSample(LiteLLMBaseModel):
     """One ``matrix`` series of the Prometheus HTTP query API."""
 
     metric: dict[str, object]
     values: list[tuple[float, str]]
 
 
-class PrometheusQueryData(BaseModel):
+class PrometheusQueryData(LiteLLMBaseModel):
     result: list[PrometheusRangeSample]
 
 
-class PrometheusQueryResponse(BaseModel):
+class PrometheusQueryResponse(LiteLLMBaseModel):
     data: PrometheusQueryData
 
 
diff --git a/litellm/integrations/shadow_eval_logger.py b/litellm/integrations/shadow_eval_logger.py
index 7f5d9fedd3d..76ba1d80841 100644
--- a/litellm/integrations/shadow_eval_logger.py
+++ b/litellm/integrations/shadow_eval_logger.py
@@ -40,6 +40,7 @@ from litellm.litellm_core_utils.llm_judge import (
 from litellm.litellm_core_utils.redact_messages import should_redact_message_logging
 from litellm.llms.base_llm.base_utils import type_to_response_format_param
 from litellm.router_utils.common_utils import resolve_model_group_alias
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.management_endpoints.auto_router_endpoints import ShadowEvalDirection
 from litellm.types.utils import SHADOW_EVAL_JUDGE_CALL_ORIGIN, SHADOW_EVAL_ROUTER_CALL_ORIGIN
 
@@ -467,7 +468,7 @@ Return ONLY valid JSON in this exact format, no other text:
 }"""
 
 
-class PairwiseVerdict(BaseModel):
+class PairwiseVerdict(LiteLLMBaseModel):
     """The judge's blind A/B verdict: the response_format schema sent with the judge call
     and the validation contract on its reply. Both fields are required and preference is
     closed over the prompt's labels, so a malformed or truncated reply is an
@@ -731,7 +732,7 @@ class _JudgeVerdict:
     cost: float
 
 
-class ActiveShadowEvalJob(BaseModel):
+class ActiveShadowEvalJob(LiteLLMBaseModel):
     """One active job as the sampling path needs it, validated straight off the untyped
     job row: immutable config plus the attempt count as of the cache fill (the turn
     budget's staleness is bounded by the cache TTL). Every way a row can be unsamplable
diff --git a/litellm/integrations/zerobus/client.py b/litellm/integrations/zerobus/client.py
index bf3a9e3e269..b803b9d63bb 100644
--- a/litellm/integrations/zerobus/client.py
+++ b/litellm/integrations/zerobus/client.py
@@ -14,7 +14,7 @@ from collections.abc import Callable, Mapping, Sequence
 from typing import Final
 
 import httpx
-from pydantic import BaseModel, ValidationError
+from pydantic import ValidationError
 
 import litellm
 from litellm.llms.custom_httpx.http_handler import AsyncHTTPHandler
@@ -25,12 +25,13 @@ from litellm.types.integrations.zerobus import (
     ZerobusConnection,
     ZerobusIngestFailure,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 TOKEN_PATH: Final = "/oidc/v1/token"
 OAUTH_SCOPE: Final = "all-apis"
 
 
-class _TokenResponse(BaseModel):
+class _TokenResponse(LiteLLMBaseModel):
     access_token: str
     expires_in: float = 3600
 
diff --git a/litellm/interactions/background_cost_polling.py b/litellm/interactions/background_cost_polling.py
index ca93436490d..21d438d9621 100644
--- a/litellm/interactions/background_cost_polling.py
+++ b/litellm/interactions/background_cost_polling.py
@@ -45,7 +45,7 @@ from datetime import datetime, timezone
 from types import MappingProxyType
 from typing import TYPE_CHECKING, Final, Literal, Protocol, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, JsonValue, TypeAdapter, ValidationError
+from pydantic import ConfigDict, JsonValue, TypeAdapter, ValidationError
 from pydantic_core import PydanticSerializationError, to_jsonable_python
 
 from litellm._logging import verbose_logger
@@ -57,6 +57,7 @@ from litellm.constants import (
 )
 from litellm.litellm_core_utils.core_helpers import get_litellm_metadata_from_kwargs
 from litellm.types.interactions import InteractionsAPIResponse
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import CustomPricingLiteLLMParams
 
 if TYPE_CHECKING:
@@ -73,7 +74,7 @@ _STATUSES_THAT_PRODUCED_OUTPUT: Final = frozenset({"completed", "requires_action
 SettlementOutcome: TypeAlias = Literal["billed", "released", "unsettled"]
 
 
-class BackgroundInteractionCreateContext(BaseModel):
+class BackgroundInteractionCreateContext(LiteLLMBaseModel):
     """
     The part of a create's logging state that billing its settled result needs,
     in a shape any replica can store and rebuild a logging object from. Provider
diff --git a/litellm/litellm_core_utils/audio_utils/subtitle_utils.py b/litellm/litellm_core_utils/audio_utils/subtitle_utils.py
index 91427ba09ad..34eee885911 100644
--- a/litellm/litellm_core_utils/audio_utils/subtitle_utils.py
+++ b/litellm/litellm_core_utils/audio_utils/subtitle_utils.py
@@ -6,7 +6,9 @@ from dataclasses import dataclass
 from itertools import accumulate, groupby
 from typing import Final
 
-from pydantic import BaseModel, ConfigDict, TypeAdapter, ValidationError
+from pydantic import ConfigDict, TypeAdapter, ValidationError
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 CUE_MAX_CHARS: Final = 84
 CUE_MAX_DURATION_MS: Final = 7000
@@ -230,7 +232,7 @@ def render_subtitle_tokens_as_vtt(tokens: Sequence[SubtitleToken]) -> str:
     return _render_vtt(group_subtitle_tokens_into_cues(tokens))
 
 
-class TranscriptionWordTiming(BaseModel):
+class TranscriptionWordTiming(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     word: str = ""
diff --git a/litellm/litellm_core_utils/cli_token_utils.py b/litellm/litellm_core_utils/cli_token_utils.py
index ee506a69ef9..ab386e10951 100644
--- a/litellm/litellm_core_utils/cli_token_utils.py
+++ b/litellm/litellm_core_utils/cli_token_utils.py
@@ -20,7 +20,7 @@ from pathlib import Path
 from types import MappingProxyType
 from typing import Final, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, ValidationError
+from pydantic import ConfigDict, ValidationError
 
 from litellm.litellm_core_utils.cli_keyring import (
     SYSTEM_KEYRING,
@@ -44,6 +44,7 @@ from litellm.litellm_core_utils.private_json import (
     stage_private_json,
     write_private_json,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
 @dataclass(frozen=True, slots=True)
@@ -83,7 +84,7 @@ SecretSave: TypeAlias = SecretWrite | CredentialNotSaved | CredentialNotRecorded
 SecretClear: TypeAlias = SecretErase | CredentialNotCleared
 
 
-class CliTokenRecord(BaseModel):
+class CliTokenRecord(LiteLLMBaseModel):
     """A stored CLI credential.
 
     `key is None` means the metadata was found but the secret could not be
@@ -104,7 +105,7 @@ class CliTokenRecord(BaseModel):
     refresh_token: str | None = None
 
 
-class CliTokenSecret(BaseModel):
+class CliTokenSecret(LiteLLMBaseModel):
     """The secret material as stored in the OS keychain.
 
     `base_url` is duplicated from the metadata file purely as a pairing tag: a
diff --git a/litellm/litellm_core_utils/get_blog_posts.py b/litellm/litellm_core_utils/get_blog_posts.py
index 25a5c7cd9ab..276d2a968a1 100644
--- a/litellm/litellm_core_utils/get_blog_posts.py
+++ b/litellm/litellm_core_utils/get_blog_posts.py
@@ -17,21 +17,21 @@ from importlib.resources import files
 from typing import Final
 
 import httpx
-from pydantic import BaseModel
 
 from litellm import verbose_logger
+from litellm.types.llms.base import LiteLLMBaseModel
 
 BLOG_POSTS_TTL_SECONDS: Final[int] = 3600  # 1 hour
 
 
-class BlogPost(BaseModel):
+class BlogPost(LiteLLMBaseModel):
     title: str
     description: str
     date: str
     url: str
 
 
-class BlogPostsResponse(BaseModel):
+class BlogPostsResponse(LiteLLMBaseModel):
     posts: list[BlogPost]
 
 
diff --git a/litellm/litellm_core_utils/llm_cost_calc/guardrail_cost.py b/litellm/litellm_core_utils/llm_cost_calc/guardrail_cost.py
index 19adf1a30a7..61ec2fc39c1 100644
--- a/litellm/litellm_core_utils/llm_cost_calc/guardrail_cost.py
+++ b/litellm/litellm_core_utils/llm_cost_calc/guardrail_cost.py
@@ -2,22 +2,23 @@ import math
 from collections.abc import Mapping
 from typing import Annotated, Final
 
-from pydantic import BaseModel, ConfigDict, Field, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, TypeAdapter, ValidationError
 
 import litellm
 from litellm._logging import verbose_logger
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import CostBreakdown
 
 BEDROCK_GUARDRAIL_PRICING_KEY: Final = "bedrock/guardrails"
 
 
-class GuardrailPricing(BaseModel):
+class GuardrailPricing(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     guardrail_cost_per_unit: Mapping[str, float]
 
 
-class GuardrailCostEntry(BaseModel):
+class GuardrailCostEntry(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     guardrail_cost: float | None = None
@@ -30,7 +31,7 @@ class GuardrailCostEntry(BaseModel):
 _GUARDRAIL_COST_ENTRY_ADAPTER: Final[TypeAdapter[GuardrailCostEntry]] = TypeAdapter(GuardrailCostEntry)
 
 
-class GuardrailCostByUnitEntry(BaseModel):
+class GuardrailCostByUnitEntry(LiteLLMBaseModel):
     """The rollup-side view of a ``guardrail_information`` entry, validated apart from
     ``GuardrailCostEntry`` so a forged per-counter map can never zero the spend path."""
 
diff --git a/litellm/litellm_core_utils/prompt_templates/factory.py b/litellm/litellm_core_utils/prompt_templates/factory.py
index 90a60a9e46c..4f6d9af487b 100644
--- a/litellm/litellm_core_utils/prompt_templates/factory.py
+++ b/litellm/litellm_core_utils/prompt_templates/factory.py
@@ -12,6 +12,7 @@ from types import MappingProxyType
 from typing import Any, Final, TypeAlias, TypedDict, cast, overload
 
 from jinja2.sandbox import ImmutableSandboxedEnvironment
+from pydantic import BaseModel
 
 import litellm
 import litellm.types
diff --git a/litellm/litellm_core_utils/served_output_texts.py b/litellm/litellm_core_utils/served_output_texts.py
index 323561a3dd2..2788720448b 100644
--- a/litellm/litellm_core_utils/served_output_texts.py
+++ b/litellm/litellm_core_utils/served_output_texts.py
@@ -6,8 +6,9 @@ from __future__ import annotations
 from collections.abc import Sequence
 from typing import Final, Literal
 
-from pydantic import BaseModel, TypeAdapter, ValidationError
+from pydantic import BaseModel, Field, TypeAdapter, ValidationError
 
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import ModelResponse, ModelResponseStream
 
 SERVED_OUTPUT_TEXTS_KEY: Final = "served_output_texts"
@@ -19,27 +20,27 @@ _TEXTS: Final = TypeAdapter(tuple[str | None, ...])
 ServedTexts = tuple[str | None, ...]
 
 
-class _TextBlock(BaseModel):
+class _TextBlock(LiteLLMBaseModel):
     type: str
     text: str | None = None
 
 
-class _AnthropicMessage(BaseModel):
+class _AnthropicMessage(LiteLLMBaseModel):
     type: Literal["message"]
     content: list[_TextBlock]
 
 
-class _ResponsesOutputItem(BaseModel):
+class _ResponsesOutputItem(LiteLLMBaseModel):
     type: str
-    content: list[_TextBlock] = []
+    content: list[_TextBlock] = Field(default=[])
 
 
-class _ResponsesResponse(BaseModel):
+class _ResponsesResponse(LiteLLMBaseModel):
     object: Literal["response"]
     output: list[_ResponsesOutputItem]
 
 
-class _ChatChoices(BaseModel):
+class _ChatChoices(LiteLLMBaseModel):
     choices: list[object]
 
 
diff --git a/litellm/litellm_core_utils/streaming_handler.py b/litellm/litellm_core_utils/streaming_handler.py
index 9d33f86d841..6080d52ea89 100644
--- a/litellm/litellm_core_utils/streaming_handler.py
+++ b/litellm/litellm_core_utils/streaming_handler.py
@@ -25,6 +25,7 @@ from litellm.litellm_core_utils.model_response_utils import (
 )
 from litellm.litellm_core_utils.redact_messages import LiteLLMLoggingObject
 from litellm.litellm_core_utils.thread_pool_executor import executor
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import OpenAIChatCompletionChunk
 from litellm.types.router import GenericLiteLLMParams
 from litellm.types.utils import (
@@ -164,7 +165,7 @@ class _VertexChunkLike(Protocol):
     candidates: Sequence[_VertexCandidateLike]
 
 
-class _ParsedChunkHiddenParams(BaseModel):
+class _ParsedChunkHiddenParams(LiteLLMBaseModel):
     provider_specific_fields: Mapping[str, object] | None = None
 
 
diff --git a/litellm/llms/anthropic/cache_aware_routing.py b/litellm/llms/anthropic/cache_aware_routing.py
index e1a50781ace..024631c961a 100644
--- a/litellm/llms/anthropic/cache_aware_routing.py
+++ b/litellm/llms/anthropic/cache_aware_routing.py
@@ -7,7 +7,7 @@ from types import MappingProxyType
 from typing import Final
 from urllib.parse import urlparse
 
-from pydantic import BaseModel, JsonValue, TypeAdapter
+from pydantic import JsonValue, TypeAdapter
 
 import litellm
 from litellm._internal_context import current_billing_time, pinned_billing_time
@@ -25,6 +25,7 @@ from litellm.llms.anthropic.prompt_cache_prediction import (
 )
 from litellm.proxy.common_utils.prompt_cache_pricing import price_cache_tokens
 from litellm.proxy.hooks.prompt_cache_prediction import lookup
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.management_endpoints.prompt_cache_prediction import (
     CacheCostScenario,
     CacheEvidence,
@@ -57,7 +58,7 @@ _NATIVE_OPTIONS: Final = frozenset(
 )
 
 
-class _ModelLimits(BaseModel):
+class _ModelLimits(LiteLLMBaseModel):
     max_input_tokens: int | None = None
     max_output_tokens: int | None = None
 
diff --git a/litellm/llms/anthropic/common_utils.py b/litellm/llms/anthropic/common_utils.py
index 476e154ed0f..049f474032c 100644
--- a/litellm/llms/anthropic/common_utils.py
+++ b/litellm/llms/anthropic/common_utils.py
@@ -12,7 +12,7 @@ from typing import Any, ClassVar, Final, Literal, TypeVar
 from urllib.parse import quote
 
 import httpx
-from pydantic import BaseModel, ConfigDict, Field, StrictBool, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, StrictBool, TypeAdapter, ValidationError
 
 import litellm
 from litellm.constants import (
@@ -51,6 +51,7 @@ from litellm.types.llms.anthropic import (
     AnthropicMessagesToolChoice,
     AnthropicThinkingParam,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import AllMessageValues
 from litellm.types.proxy.auth.special_headers import SpecialHeaders
 from litellm.types.proxy.model_listing import ModelInfoResponse
@@ -349,11 +350,11 @@ def optionally_handle_anthropic_oauth(headers: dict, api_key: str | None) -> tup
     return headers, api_key
 
 
-class _EagerInputStreamingFunction(BaseModel):
+class _EagerInputStreamingFunction(LiteLLMBaseModel):
     eager_input_streaming: StrictBool | None = None
 
 
-class _EagerInputStreamingTool(BaseModel):
+class _EagerInputStreamingTool(LiteLLMBaseModel):
     eager_input_streaming: StrictBool | None = None
     function: _EagerInputStreamingFunction | None = None
 
@@ -388,11 +389,11 @@ def _litellm_params_str(litellm_params: Mapping[str, object] | None, key: str) -
     return value if isinstance(value, str) else None
 
 
-class _AnthropicModelListEntry(BaseModel):
+class _AnthropicModelListEntry(LiteLLMBaseModel):
     id: str
 
 
-class _AnthropicModelsPage(BaseModel):
+class _AnthropicModelsPage(LiteLLMBaseModel):
     data: Sequence[_AnthropicModelListEntry] = Field(default_factory=tuple)
     has_more: bool = False
     last_id: str | None = None
@@ -1784,13 +1785,13 @@ def sanitize_tool_use_ids_in_anthropic_messages(messages: list[Any]) -> list[Any
     return out
 
 
-class _ReplayedSearchQuery(BaseModel):
+class _ReplayedSearchQuery(LiteLLMBaseModel):
     model_config = ConfigDict(extra="allow")
 
     query: str = ""
 
 
-class _ReplayedWebSearchResult(BaseModel):
+class _ReplayedWebSearchResult(LiteLLMBaseModel):
     model_config = ConfigDict(extra="allow")
 
     type: Literal["web_search_result"]
@@ -1800,14 +1801,14 @@ class _ReplayedWebSearchResult(BaseModel):
     encrypted_content: str = ""
 
 
-class _ReplayedWebSearchToolResultError(BaseModel):
+class _ReplayedWebSearchToolResultError(LiteLLMBaseModel):
     model_config = ConfigDict(extra="allow")
 
     type: Literal["web_search_tool_result_error"]
     error_code: str = ""
 
 
-class _ReplayedWebSearchToolResult(BaseModel):
+class _ReplayedWebSearchToolResult(LiteLLMBaseModel):
     model_config = ConfigDict(extra="allow")
 
     type: Literal["web_search_tool_result"]
@@ -1815,7 +1816,7 @@ class _ReplayedWebSearchToolResult(BaseModel):
     content: tuple[_ReplayedWebSearchResult, ...] | _ReplayedWebSearchToolResultError
 
 
-class _ReplayedServerToolUse(BaseModel):
+class _ReplayedServerToolUse(LiteLLMBaseModel):
     model_config = ConfigDict(extra="allow")
 
     type: Literal["server_tool_use"]
@@ -1823,7 +1824,7 @@ class _ReplayedServerToolUse(BaseModel):
     input: _ReplayedSearchQuery = _ReplayedSearchQuery()
 
 
-class _TextBlock(BaseModel):
+class _TextBlock(LiteLLMBaseModel):
     type: Literal["text"] = "text"
     text: str
 
diff --git a/litellm/llms/anthropic/cost_calculation.py b/litellm/llms/anthropic/cost_calculation.py
index 4a935ac18b4..88805df2b08 100644
--- a/litellm/llms/anthropic/cost_calculation.py
+++ b/litellm/llms/anthropic/cost_calculation.py
@@ -5,13 +5,14 @@ Helper util for handling anthropic-specific cost calculation
 
 from typing import TYPE_CHECKING, Final, Optional
 
-from pydantic import BaseModel, ValidationError
+from pydantic import ValidationError
 
 from litellm.litellm_core_utils.llm_cost_calc.utils import (
     generic_cost_per_token,
     get_provider_specific_geo_multiplier,
     get_web_search_requests_from_usage,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from litellm.types.utils import ModelInfo, Usage
@@ -71,15 +72,15 @@ def cost_per_token(
     return prompt_cost, completion_cost
 
 
-class _AnthropicServerToolUseProbe(BaseModel):
+class _AnthropicServerToolUseProbe(LiteLLMBaseModel):
     web_search_requests: int | None = None
 
 
-class _AnthropicUsageProbe(BaseModel):
+class _AnthropicUsageProbe(LiteLLMBaseModel):
     server_tool_use: _AnthropicServerToolUseProbe | None = None
 
 
-class _AnthropicResponseProbe(BaseModel):
+class _AnthropicResponseProbe(LiteLLMBaseModel):
     usage: _AnthropicUsageProbe | None = None
 
 
diff --git a/litellm/llms/anthropic/pass_through/responses_adapters/streaming_iterator.py b/litellm/llms/anthropic/pass_through/responses_adapters/streaming_iterator.py
index cc6f4da3403..450640f47b9 100644
--- a/litellm/llms/anthropic/pass_through/responses_adapters/streaming_iterator.py
+++ b/litellm/llms/anthropic/pass_through/responses_adapters/streaming_iterator.py
@@ -6,7 +6,7 @@ from collections import deque
 from collections.abc import AsyncIterator, Iterator, Mapping
 from typing import TYPE_CHECKING, Any, Final
 
-from pydantic import BaseModel, ConfigDict, field_validator
+from pydantic import ConfigDict, field_validator
 
 from litellm import verbose_logger
 from litellm._logging import redact_internal_details_from_client_message
@@ -22,6 +22,7 @@ from litellm.llms.anthropic.pass_through.messages.utils import (
 )
 from litellm.responses.streaming_iterator import stream_error_status_and_message
 from litellm.types.llms.anthropic_messages.anthropic_response import AnthropicUsage
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .transformation import (
     REASONING_SUMMARY_PART_SEPARATOR,
@@ -32,7 +33,7 @@ if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObject
 
 
-class _UpstreamFailure(BaseModel):
+class _UpstreamFailure(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     status_code: int | None = None
@@ -56,13 +57,13 @@ class _UpstreamFailure(BaseModel):
         return value if isinstance(value, str) else None
 
 
-class _FailedResponse(BaseModel):
+class _FailedResponse(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, from_attributes=True)
 
     error: object | None = None
 
 
-class _FailedResponseEvent(BaseModel):
+class _FailedResponseEvent(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, from_attributes=True)
 
     response: _FailedResponse | None = None
diff --git a/litellm/llms/anthropic/pass_through/utils.py b/litellm/llms/anthropic/pass_through/utils.py
index 335a0e5641d..95baecbfbfe 100644
--- a/litellm/llms/anthropic/pass_through/utils.py
+++ b/litellm/llms/anthropic/pass_through/utils.py
@@ -3,9 +3,10 @@ from collections.abc import Mapping
 from types import MappingProxyType
 from typing import TYPE_CHECKING, Final
 
-from pydantic import BaseModel, ConfigDict, ValidationError
+from pydantic import ConfigDict, ValidationError
 
 import litellm
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import ModelInfo
 
 if TYPE_CHECKING:
@@ -23,7 +24,7 @@ _EFFORT_DEGRADATION_CHAIN: Final[Mapping[str, tuple[str, ...]]] = MappingProxyTy
 _THINKING_OFF: Final = "none"
 
 
-class _ClaudeCodeUserId(BaseModel):
+class _ClaudeCodeUserId(LiteLLMBaseModel):
     """The JSON Claude Code packs into ``metadata.user_id``; only ``session_id`` is per conversation."""
 
     model_config = ConfigDict(frozen=True)
diff --git a/litellm/llms/anthropic/prompt_cache_prediction.py b/litellm/llms/anthropic/prompt_cache_prediction.py
index ff304e845f7..00fe56a5e39 100644
--- a/litellm/llms/anthropic/prompt_cache_prediction.py
+++ b/litellm/llms/anthropic/prompt_cache_prediction.py
@@ -10,7 +10,7 @@ from types import MappingProxyType
 from typing import Annotated, Final, Literal, Protocol, TypeAlias
 
 import httpx
-from pydantic import BaseModel, ConfigDict, Field, JsonValue, StrictInt, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, JsonValue, StrictInt, TypeAdapter, ValidationError
 
 import litellm
 from litellm.llms.anthropic.common_utils import AnthropicModelInfo, is_anthropic_oauth_key
@@ -20,6 +20,7 @@ from litellm.llms.anthropic.pass_through.messages.transformation import (
     DEFAULT_ANTHROPIC_API_VERSION,
     AnthropicMessagesConfig,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.router import LiteLLM_Params
 from litellm.types.utils import ModelResponse
 from litellm.utils import supports_thinking_cache_preservation
@@ -65,7 +66,7 @@ _DEPLOYMENT_OPTIONS: Final = frozenset(
 )
 
 
-class _StrictModel(BaseModel):
+class _StrictModel(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True, strict=True)
 
 
@@ -455,37 +456,37 @@ def cache_scope(
     return _digest((caller_key_hash, deployment_id, provider_key, model, anthropic_version))
 
 
-class _TTLUsage(BaseModel):
+class _TTLUsage(LiteLLMBaseModel):
     model_config = ConfigDict(strict=True)
     ephemeral_5m_input_tokens: int = Field(default=0, ge=0)
     ephemeral_1h_input_tokens: int = Field(default=0, ge=0)
 
 
-class _CacheUsage(BaseModel):
+class _CacheUsage(LiteLLMBaseModel):
     model_config = ConfigDict(strict=True)
     cached_tokens: int = Field(default=0, ge=0)
     cache_creation_tokens: int = Field(default=0, ge=0)
     cache_creation_token_details: _TTLUsage | None = None
 
 
-class _Usage(BaseModel):
+class _Usage(LiteLLMBaseModel):
     model_config = ConfigDict(strict=True)
     prompt_tokens: int = Field(ge=0)
     prompt_tokens_details: _CacheUsage
 
 
-class _Choice(BaseModel):
+class _Choice(LiteLLMBaseModel):
     finish_reason: str = Field(min_length=1)
 
 
-class _Response(BaseModel):
+class _Response(LiteLLMBaseModel):
     model_config = ConfigDict(strict=True)
     model: str
     usage: _Usage
     choices: tuple[_Choice, ...] = Field(min_length=1, strict=False)
 
 
-class _CountBody(BaseModel):
+class _CountBody(LiteLLMBaseModel):
     messages: Sequence[Mapping[str, JsonValue]]
     tools: Sequence[Mapping[str, JsonValue]] | None = None
     system: str | Sequence[Mapping[str, JsonValue]] | None = None
@@ -494,7 +495,7 @@ class _CountBody(BaseModel):
     output_config: Mapping[str, JsonValue] | None = None
 
 
-class _CountResult(BaseModel):
+class _CountResult(LiteLLMBaseModel):
     input_tokens: Annotated[StrictInt, Field(ge=0)]
 
 
diff --git a/litellm/llms/anthropic/wif.py b/litellm/llms/anthropic/wif.py
index b096a11ace4..2444a8f833e 100644
--- a/litellm/llms/anthropic/wif.py
+++ b/litellm/llms/anthropic/wif.py
@@ -10,7 +10,7 @@ from types import MappingProxyType
 from typing import Final, NoReturn, TypeVar
 from urllib.parse import urlsplit, urlunsplit
 
-from pydantic import BaseModel, ConfigDict, ValidationError
+from pydantic import ConfigDict, ValidationError
 from typing_extensions import assert_never
 
 import litellm
@@ -42,6 +42,7 @@ from litellm.llms.base_llm.auth.types import (
     TokenTransportError,
 )
 from litellm.types.llms.anthropic import ANTHROPIC_TOKEN_EXCHANGE_PATH
+from litellm.types.llms.base import LiteLLMBaseModel
 
 _JWT_BEARER_GRANT_TYPE: Final = "urn:ietf:params:oauth:grant-type:jwt-bearer"
 _DEFAULT_API_BASE: Final = "https://api.anthropic.com"
@@ -119,7 +120,7 @@ _EMPTY_PARAMS: Final[Mapping[str, object]] = MappingProxyType({})
 _IdentitySourceVariant = TypeVar("_IdentitySourceVariant", bound="InternalIssuerSource | KeycloakSource")
 
 
-class AnthropicWifParams(BaseModel):
+class AnthropicWifParams(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     federation_rule_id: str
diff --git a/litellm/llms/azure/passthrough/transformation.py b/litellm/llms/azure/passthrough/transformation.py
index 8ea0ead2e91..ff83151626a 100644
--- a/litellm/llms/azure/passthrough/transformation.py
+++ b/litellm/llms/azure/passthrough/transformation.py
@@ -5,7 +5,7 @@ from typing import TYPE_CHECKING, Final, Optional
 
 import httpx
 from httpx import Response
-from pydantic import BaseModel, ValidationError
+from pydantic import ValidationError
 
 from litellm.litellm_core_utils.litellm_logging import Logging
 from litellm.llms.azure.common_utils import BaseAzureLLM
@@ -17,6 +17,7 @@ from litellm.llms.base_llm.passthrough.transformation import (
     strip_leading_model_segment,
 )
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import AllMessageValues, ResponsesAPIResponse, ResponsesTerminalEvent
 from litellm.types.router import GenericLiteLLMParams
 from litellm.types.utils import CallTypes, EmbeddingResponse, ImageResponse
@@ -27,11 +28,11 @@ if TYPE_CHECKING:
     from litellm.llms.base_llm.passthrough.transformation import LoggedRelayResponse
 
 
-class RelayedChatRequest(BaseModel):
+class RelayedChatRequest(LiteLLMBaseModel):
     messages: Sequence[Mapping[str, object]] | None = None
 
 
-class RelayedCallDetails(BaseModel):
+class RelayedCallDetails(LiteLLMBaseModel):
     request_data: RelayedChatRequest | None = None
 
 
diff --git a/litellm/llms/azure/search/transformation.py b/litellm/llms/azure/search/transformation.py
index 45ad78df687..185aedc4ad6 100644
--- a/litellm/llms/azure/search/transformation.py
+++ b/litellm/llms/azure/search/transformation.py
@@ -33,7 +33,7 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Final, Literal
 
 import httpx
-from pydantic import BaseModel, ConfigDict, ValidationError
+from pydantic import ConfigDict, ValidationError
 
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
 from litellm.llms.base_llm.search.transformation import (
@@ -42,6 +42,7 @@ from litellm.llms.base_llm.search.transformation import (
     SearchResult,
 )
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
@@ -61,7 +62,7 @@ _UPSTREAM_ERROR_STATUS: Final = 502
 _RESPONSE_COST_HEADER: Final = "llm_provider-x-litellm-response-cost"
 
 
-class _Annotation(BaseModel):
+class _Annotation(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     type: str = ""
@@ -71,7 +72,7 @@ class _Annotation(BaseModel):
     end_index: int | None = None
 
 
-class _ContentPart(BaseModel):
+class _ContentPart(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     type: str = ""
@@ -79,26 +80,26 @@ class _ContentPart(BaseModel):
     annotations: tuple[_Annotation, ...] = ()
 
 
-class _OutputItem(BaseModel):
+class _OutputItem(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     type: str = ""
     content: tuple[_ContentPart, ...] = ()
 
 
-class _ErrorBody(BaseModel):
+class _ErrorBody(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     message: str | None = None
 
 
-class _IncompleteDetails(BaseModel):
+class _IncompleteDetails(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     reason: str | None = None
 
 
-class _ResponsesEnvelope(BaseModel):
+class _ResponsesEnvelope(LiteLLMBaseModel):
     """A Foundry Responses API body. `output` is required: a body without it is not a
     Responses API response and must not be reported as a successful empty search.
 
@@ -113,7 +114,7 @@ class _ResponsesEnvelope(BaseModel):
     incomplete_details: _IncompleteDetails | None = None
 
 
-class _ErrorEnvelope(BaseModel):
+class _ErrorEnvelope(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     error: _ErrorBody | None = None
@@ -205,41 +206,41 @@ def _capped(results: tuple[SearchResult, ...], max_results: int | None) -> tuple
     return results[:max_results] if max_results is not None else results
 
 
-class _SearchConfiguration(BaseModel):
+class _SearchConfiguration(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     project_connection_id: str
     count: int | None = None
 
 
-class _BingGroundingParams(BaseModel):
+class _BingGroundingParams(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     search_configurations: tuple[_SearchConfiguration, ...]
 
 
-class _BingGroundingTool(BaseModel):
+class _BingGroundingTool(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     type: Literal["bing_grounding"] = "bing_grounding"
     bing_grounding: _BingGroundingParams
 
 
-class _UserLocation(BaseModel):
+class _UserLocation(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     type: Literal["approximate"] = "approximate"
     country: str
 
 
-class _WebSearchTool(BaseModel):
+class _WebSearchTool(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     type: Literal["web_search"] = "web_search"
     user_location: _UserLocation | None = None
 
 
-class _ResponsesRequest(BaseModel):
+class _ResponsesRequest(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     model: str
diff --git a/litellm/llms/base_llm/auth/client_credentials.py b/litellm/llms/base_llm/auth/client_credentials.py
index bf86dc6508f..715eda7db89 100644
--- a/litellm/llms/base_llm/auth/client_credentials.py
+++ b/litellm/llms/base_llm/auth/client_credentials.py
@@ -18,7 +18,7 @@ from typing import TYPE_CHECKING, Final, TypeAlias
 from urllib.parse import quote, quote_plus, urlencode
 
 import httpx
-from pydantic import BaseModel, SecretStr, ValidationError
+from pydantic import SecretStr, ValidationError
 from typing_extensions import assert_never
 
 from litellm.llms.base_llm.auth.identity_source import KeycloakSource, ref_for_error_message
@@ -30,6 +30,7 @@ from litellm.llms.base_llm.auth.token_exchange import (
     validate_token_endpoint_url,
 )
 from litellm.llms.base_llm.auth.types import InsecureTokenUrl, SyncTokenPoster
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from litellm.llms.custom_httpx.http_handler import HTTPHandler
@@ -41,7 +42,7 @@ _TIMEOUT_SECONDS: Final = 30.0
 _FORM_CONTENT_TYPE: Final = "application/x-www-form-urlencoded"
 
 
-class _ClientCredentialsResponse(BaseModel):
+class _ClientCredentialsResponse(LiteLLMBaseModel):
     access_token: str
 
 
diff --git a/litellm/llms/base_llm/auth/identity_source.py b/litellm/llms/base_llm/auth/identity_source.py
index f8f02249b3c..0898b7b0b8b 100644
--- a/litellm/llms/base_llm/auth/identity_source.py
+++ b/litellm/llms/base_llm/auth/identity_source.py
@@ -11,7 +11,9 @@ import hashlib
 from enum import Enum
 from typing import Annotated, Final, Literal, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, Field, TypeAdapter
+from pydantic import ConfigDict, Field, TypeAdapter
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 _REF_HASH_HEX_LENGTH: Final = 16
 _MAX_TTL_SECONDS: Final = 3600
@@ -23,7 +25,7 @@ class AnthropicIdentitySourceKind(str, Enum):
     keycloak = "keycloak"
 
 
-class InternalIssuerSource(BaseModel):
+class InternalIssuerSource(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="forbid", hide_input_in_errors=True)
 
     kind: Literal[AnthropicIdentitySourceKind.internal_issuer] = AnthropicIdentitySourceKind.internal_issuer
@@ -34,7 +36,7 @@ class InternalIssuerSource(BaseModel):
     signing_key_ref: str
 
 
-class KeycloakSource(BaseModel):
+class KeycloakSource(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="forbid", hide_input_in_errors=True)
 
     kind: Literal[AnthropicIdentitySourceKind.keycloak] = AnthropicIdentitySourceKind.keycloak
diff --git a/litellm/llms/base_llm/auth/shared_token_store.py b/litellm/llms/base_llm/auth/shared_token_store.py
index 878df1dc6fb..772a9a2ad9e 100644
--- a/litellm/llms/base_llm/auth/shared_token_store.py
+++ b/litellm/llms/base_llm/auth/shared_token_store.py
@@ -16,9 +16,10 @@ from dataclasses import dataclass
 from pathlib import Path
 from typing import Final, Protocol
 
-from pydantic import BaseModel, SecretStr, ValidationError
+from pydantic import SecretStr, ValidationError
 
 from litellm._logging import verbose_logger
+from litellm.types.llms.base import LiteLLMBaseModel
 
 CACHE_DIR_ENV: Final = "LITELLM_TOKEN_EXCHANGE_CACHE_DIR"
 
@@ -43,7 +44,7 @@ class SharedTokenStore(Protocol):
     def lock(self, key: str) -> contextlib.AbstractContextManager[None]: ...
 
 
-class _StoredTokenFile(BaseModel):
+class _StoredTokenFile(LiteLLMBaseModel):
     access_token: str
     expires_at_epoch: float | None
     assertion_sha256: str
diff --git a/litellm/llms/base_llm/auth/token_exchange.py b/litellm/llms/base_llm/auth/token_exchange.py
index 5f4b8c00bba..5bf1f4dd0db 100644
--- a/litellm/llms/base_llm/auth/token_exchange.py
+++ b/litellm/llms/base_llm/auth/token_exchange.py
@@ -23,7 +23,7 @@ from typing import TYPE_CHECKING, Final, Protocol, TypeAlias
 from urllib.parse import unquote, unquote_plus, urlencode, urlsplit, urlunsplit
 
 import httpx
-from pydantic import BaseModel, SecretStr, TypeAdapter, ValidationError
+from pydantic import SecretStr, TypeAdapter, ValidationError
 from typing_extensions import assert_never
 
 from litellm._logging import verbose_logger
@@ -44,6 +44,7 @@ from litellm.llms.base_llm.auth.types import (
     TokenExchangeSpec,
     TokenTransportError,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.services import ServiceTypes
 
 if TYPE_CHECKING:
@@ -86,7 +87,7 @@ _CREDENTIAL_CHARS: Final = re.compile(r"[^A-Za-z0-9._~+/=-]")
 _SENTINEL_BODY_MESSAGES: Final = frozenset({_OVERSIZED_BODY_MESSAGE, _NON_OBJECT_BODY_MESSAGE})
 
 
-class _TokenExchangeResponse(BaseModel):
+class _TokenExchangeResponse(LiteLLMBaseModel):
     access_token: str
     expires_in: int | None = None
     token_type: str | None = None
diff --git a/litellm/llms/base_llm/passthrough/transformation.py b/litellm/llms/base_llm/passthrough/transformation.py
index 84cbd4204e3..f75f9159070 100644
--- a/litellm/llms/base_llm/passthrough/transformation.py
+++ b/litellm/llms/base_llm/passthrough/transformation.py
@@ -6,8 +6,9 @@ from collections.abc import Callable, Mapping, Sequence
 from dataclasses import dataclass
 from typing import TYPE_CHECKING, Final, Protocol, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, TypeAdapter, ValidationError
+from pydantic import ConfigDict, TypeAdapter, ValidationError
 
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import CallTypes
 
 from ..base_utils import BaseLLMModelInfo
@@ -29,7 +30,7 @@ if TYPE_CHECKING:
 RELAYED_JSON_OBJECT: Final = TypeAdapter(Mapping[str, object])
 
 
-class PassthroughMetadata(BaseModel):
+class PassthroughMetadata(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     model_group: str = ""
diff --git a/litellm/llms/bedrock/base_aws_llm.py b/litellm/llms/bedrock/base_aws_llm.py
index badb76d00c7..626b0d78589 100644
--- a/litellm/llms/bedrock/base_aws_llm.py
+++ b/litellm/llms/bedrock/base_aws_llm.py
@@ -15,7 +15,7 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Any, ClassVar, Final, Literal, ParamSpec, TypeVar, cast, get_args, overload
 
 import httpx
-from pydantic import BaseModel, TypeAdapter, ValidationError
+from pydantic import TypeAdapter, ValidationError
 from typing_extensions import NotRequired, ReadOnly, TypedDict
 
 from litellm._logging import verbose_logger
@@ -33,6 +33,7 @@ from litellm.constants import (
 from litellm.litellm_core_utils.aws_partition import contains_bedrock_arn, get_aws_dns_suffix
 from litellm.litellm_core_utils.dd_tracing import tracer
 from litellm.secret_managers.main import get_secret, get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.bedrock import AWS_AUTH_PARAM_KEYS, AwsAuthParams, AwsSessionTag
 
 if TYPE_CHECKING:
@@ -176,7 +177,7 @@ def pop_aws_auth_params(
     )
 
 
-class BedrockRequestTarget(BaseModel):
+class BedrockRequestTarget(LiteLLMBaseModel):
     aws_region_name: str
     aws_bedrock_runtime_endpoint: str | None
 
@@ -194,7 +195,7 @@ def bedrock_bearer_token(api_key: str | None) -> str | None:
     return token or None
 
 
-class _WebIdentityTokenClaims(BaseModel):
+class _WebIdentityTokenClaims(LiteLLMBaseModel):
     aud: str | list[str] | None = None
     iss: str | None = None
 
diff --git a/litellm/llms/bedrock/embed/twelvelabs_marengo_3_transformation.py b/litellm/llms/bedrock/embed/twelvelabs_marengo_3_transformation.py
index 4aac6f22bae..44084e30961 100644
--- a/litellm/llms/bedrock/embed/twelvelabs_marengo_3_transformation.py
+++ b/litellm/llms/bedrock/embed/twelvelabs_marengo_3_transformation.py
@@ -9,10 +9,11 @@ from collections.abc import Mapping
 from types import MappingProxyType
 from typing import Final
 
-from pydantic import BaseModel, ConfigDict, TypeAdapter, ValidationError
+from pydantic import ConfigDict, TypeAdapter, ValidationError
 from typing_extensions import assert_never
 
 from litellm.llms.bedrock.common_utils import BedrockError
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.bedrock import (
     TWELVELABS_MARENGO_3_EMBEDDING_OPTIONS,
     TWELVELABS_MARENGO_3_EMBEDDING_SCOPES,
@@ -57,7 +58,7 @@ def is_marengo_3_model(model: str | None) -> bool:
     return MARENGO_3_MODEL_MARKER in (model or "")
 
 
-class Marengo3Params(BaseModel):
+class Marengo3Params(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     inputType: TWELVELABS_MARENGO_3_INPUT_TYPES | None = None
diff --git a/litellm/llms/bedrock/embed/twelvelabs_marengo_transformation.py b/litellm/llms/bedrock/embed/twelvelabs_marengo_transformation.py
index 65ca2be191f..9391068b8aa 100644
--- a/litellm/llms/bedrock/embed/twelvelabs_marengo_transformation.py
+++ b/litellm/llms/bedrock/embed/twelvelabs_marengo_transformation.py
@@ -10,7 +10,7 @@ Marengo 3.0 docs - https://docs.aws.amazon.com/bedrock/latest/userguide/model-pa
 from collections.abc import Mapping
 from typing import Final, cast
 
-from pydantic import BaseModel, ConfigDict, TypeAdapter
+from pydantic import ConfigDict, TypeAdapter
 from typing_extensions import assert_never
 
 import litellm
@@ -19,6 +19,7 @@ from litellm.llms.bedrock.embed.twelvelabs_marengo_3_transformation import (
     build_marengo_3_request,
     is_marengo_3_model,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.bedrock import (
     TWELVELABS_EMBEDDING_INPUT_TYPES,
     TWELVELABS_MARENGO_3_INPUT_TYPES,
@@ -32,13 +33,13 @@ from litellm.types.llms.bedrock import (
 from litellm.types.utils import Embedding, EmbeddingResponse, PromptTokensDetailsWrapper, Usage
 
 
-class MarengoEmbeddingItem(BaseModel):
+class MarengoEmbeddingItem(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     embedding: tuple[float, ...] | None = None
 
 
-class MarengoInvokeResponse(BaseModel):
+class MarengoInvokeResponse(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     data: tuple[MarengoEmbeddingItem, ...] = ()
@@ -53,14 +54,14 @@ class MarengoInvokeResponse(BaseModel):
         return tuple(item.embedding for item in self.embeddings if item.embedding is not None)
 
 
-class MarengoBilledMultiInput(BaseModel):
+class MarengoBilledMultiInput(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     inputText: str | None = None
     mediaSources: tuple[Mapping[str, object], ...] = ()
 
 
-class MarengoBilledRequest(BaseModel):
+class MarengoBilledRequest(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     inputType: TWELVELABS_MARENGO_3_INPUT_TYPES | None = None
diff --git a/litellm/llms/bedrock/files/transformation.py b/litellm/llms/bedrock/files/transformation.py
index be6fbf6c53a..04969027638 100644
--- a/litellm/llms/bedrock/files/transformation.py
+++ b/litellm/llms/bedrock/files/transformation.py
@@ -16,7 +16,7 @@ from urllib.parse import quote, unquote, urlencode
 import httpx
 from httpx import Headers, Response
 from openai.types.file_deleted import FileDeleted
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
 from typing_extensions import ReadOnly
 
 from litellm._logging import verbose_logger
@@ -47,6 +47,7 @@ from litellm.llms.base_llm.files.transformation import (
     BaseFilesConfig,
     LiteLLMLoggingObj,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.bedrock import AwsAuthParams, BedrockBatchRecordKind
 from litellm.types.llms.openai import (
     AllMessageValues,
@@ -75,7 +76,7 @@ LIST_FILES_PURPOSE_PARAM: Final = "_s3_list_files_purpose"
 LIST_FILES_LOCATION_PARAM: Final = "_s3_list_files_location"
 
 
-class _S3DeleteContext(BaseModel):
+class _S3DeleteContext(LiteLLMBaseModel):
     file_id: str = Field(min_length=1)
 
 
@@ -154,7 +155,7 @@ class _S3RequestTarget:
     request_params: _BedrockS3RequestParams
 
 
-class _TrustedS3ModelCredentials(BaseModel):
+class _TrustedS3ModelCredentials(LiteLLMBaseModel):
     """The S3 buckets the server trusts file ids against, from the deployment snapshot."""
 
     model_config = ConfigDict(extra="ignore")
diff --git a/litellm/llms/bedrock/image_edit/handler.py b/litellm/llms/bedrock/image_edit/handler.py
index be6489f20ae..58de67107f4 100644
--- a/litellm/llms/bedrock/image_edit/handler.py
+++ b/litellm/llms/bedrock/image_edit/handler.py
@@ -10,7 +10,6 @@ import json
 from typing import TYPE_CHECKING, Any, Final
 
 import httpx
-from pydantic import BaseModel
 
 import litellm
 from litellm._logging import verbose_logger
@@ -27,6 +26,7 @@ from litellm.llms.custom_httpx.http_handler import (
     _get_httpx_client,
     get_async_httpx_client,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import ImageResponse
 
 from ..base_aws_llm import BaseAWSLLM, bedrock_bearer_token
@@ -38,7 +38,7 @@ else:
     AWSPreparedRequest = Any
 
 
-class BedrockImageEditPreparedRequest(BaseModel):
+class BedrockImageEditPreparedRequest(LiteLLMBaseModel):
     """
     Internal/Helper class for preparing the request for bedrock image edit
     """
diff --git a/litellm/llms/bedrock/image_generation/image_handler.py b/litellm/llms/bedrock/image_generation/image_handler.py
index 87762b648e0..bdf9f596053 100644
--- a/litellm/llms/bedrock/image_generation/image_handler.py
+++ b/litellm/llms/bedrock/image_generation/image_handler.py
@@ -4,7 +4,6 @@ import json
 from typing import TYPE_CHECKING, Any, Final
 
 import httpx
-from pydantic import BaseModel
 
 import litellm
 from litellm._logging import verbose_logger
@@ -27,6 +26,7 @@ from litellm.llms.custom_httpx.http_handler import (
     _get_httpx_client,
     get_async_httpx_client,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import ImageResponse
 
 from ..base_aws_llm import BaseAWSLLM, bedrock_bearer_token
@@ -38,7 +38,7 @@ else:
     AWSPreparedRequest = Any
 
 
-class BedrockImagePreparedRequest(BaseModel):
+class BedrockImagePreparedRequest(LiteLLMBaseModel):
     """
     Internal/Helper class for preparing the request for bedrock image generation
     """
diff --git a/litellm/llms/bedrock/realtime/transformation.py b/litellm/llms/bedrock/realtime/transformation.py
index 6ecbddbc558..03228fc43ef 100644
--- a/litellm/llms/bedrock/realtime/transformation.py
+++ b/litellm/llms/bedrock/realtime/transformation.py
@@ -10,7 +10,6 @@ import uuid as uuid_lib
 from typing import Final, cast
 
 import httpx
-from pydantic import BaseModel
 
 from litellm._logging import verbose_logger
 from litellm._uuid import uuid
@@ -18,6 +17,7 @@ from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLogging
 from litellm.llms.base_llm.realtime.transformation import BaseRealtimeConfig
 from litellm.llms.bedrock.common_utils import BedrockError
 from litellm.llms.bedrock.realtime.trigger_audio import ready_trigger_pcm
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import (
     OpenAIRealtimeContentPartDone,
     OpenAIRealtimeDoneEvent,
@@ -45,25 +45,25 @@ from litellm.types.realtime import (
 )
 
 
-class BedrockContentEnd(BaseModel):
+class BedrockContentEnd(LiteLLMBaseModel):
     stopReason: str | None = None
 
 
-class BedrockUsageTokenDetails(BaseModel):
+class BedrockUsageTokenDetails(LiteLLMBaseModel):
     speechTokens: int = 0
     textTokens: int = 0
 
 
-class BedrockUsageDetailsTotal(BaseModel):
+class BedrockUsageDetailsTotal(LiteLLMBaseModel):
     input: BedrockUsageTokenDetails = BedrockUsageTokenDetails()
     output: BedrockUsageTokenDetails = BedrockUsageTokenDetails()
 
 
-class BedrockUsageDetails(BaseModel):
+class BedrockUsageDetails(LiteLLMBaseModel):
     total: BedrockUsageDetailsTotal = BedrockUsageDetailsTotal()
 
 
-class BedrockUsageEvent(BaseModel):
+class BedrockUsageEvent(LiteLLMBaseModel):
     totalInputTokens: int = 0
     totalOutputTokens: int = 0
     totalTokens: int = 0
diff --git a/litellm/llms/custom_httpx/aiohttp_transport.py b/litellm/llms/custom_httpx/aiohttp_transport.py
index 73adf9c7455..b5e2f73c4cd 100644
--- a/litellm/llms/custom_httpx/aiohttp_transport.py
+++ b/litellm/llms/custom_httpx/aiohttp_transport.py
@@ -14,15 +14,16 @@ import aiohttp.client_exceptions
 import aiohttp.http_exceptions
 import httpx
 from aiohttp.client import ClientResponse, ClientSession
-from pydantic import BaseModel, TypeAdapter
+from pydantic import TypeAdapter
 from typing_extensions import ReadOnly, TypedDict
 
 import litellm
 from litellm._logging import verbose_logger
 from litellm.secret_managers.main import str_to_bool
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class HttpxTimeoutExtension(BaseModel):
+class HttpxTimeoutExtension(LiteLLMBaseModel):
     connect: float | None = None
     read: float | None = None
     write: float | None = None
diff --git a/litellm/llms/edenai/chat/transformation.py b/litellm/llms/edenai/chat/transformation.py
index 84866044103..ae5da5d1cfd 100644
--- a/litellm/llms/edenai/chat/transformation.py
+++ b/litellm/llms/edenai/chat/transformation.py
@@ -13,12 +13,13 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Final
 
 import httpx
-from pydantic import BaseModel, TypeAdapter
+from pydantic import TypeAdapter
 
 import litellm
 from litellm.litellm_core_utils.core_helpers import set_response_cost_in_hidden_params
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
 from litellm.llms.openai.chat.gpt_transformation import OpenAIChatCompletionStreamingHandler, OpenAIGPTConfig
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import AllMessageValues
 from litellm.types.utils import ModelResponse, ModelResponseStream, Usage
 
@@ -31,11 +32,11 @@ if TYPE_CHECKING:
 _OPTIONAL_MAPPING: Final[TypeAdapter[Mapping[str, object] | None]] = TypeAdapter(Mapping[str, object] | None)
 
 
-class _EdenAIModel(BaseModel):
+class _EdenAIModel(LiteLLMBaseModel):
     id: str
 
 
-class _EdenAIModelCatalog(BaseModel):
+class _EdenAIModelCatalog(LiteLLMBaseModel):
     data: tuple[_EdenAIModel, ...]
 
 
diff --git a/litellm/llms/edenai/common_utils.py b/litellm/llms/edenai/common_utils.py
index ab7ee9b1c9d..3216f798fbe 100644
--- a/litellm/llms/edenai/common_utils.py
+++ b/litellm/llms/edenai/common_utils.py
@@ -7,12 +7,13 @@ from collections.abc import Container, Mapping
 from types import MappingProxyType
 from typing import Final
 
-from pydantic import AliasChoices, BaseModel, Field, ValidationError
+from pydantic import AliasChoices, Field, ValidationError
 
 import litellm
 from litellm.exceptions import AuthenticationError
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import LlmProviders
 
 EDENAI_API_BASE: Final = "https://api.edenai.run/v3"
@@ -23,7 +24,7 @@ class EdenAIException(BaseLLMException):
     pass
 
 
-class _EdenAIExtras(BaseModel):
+class _EdenAIExtras(LiteLLMBaseModel):
     cost: float | None = Field(default=None, validation_alias=AliasChoices("cost", EDENAI_COST_HEADER))
 
 
diff --git a/litellm/llms/fal_ai/chat/transformation.py b/litellm/llms/fal_ai/chat/transformation.py
index d107426d793..115b1f840de 100644
--- a/litellm/llms/fal_ai/chat/transformation.py
+++ b/litellm/llms/fal_ai/chat/transformation.py
@@ -10,11 +10,12 @@ from collections.abc import Mapping, Sequence
 from typing import TYPE_CHECKING, Final
 
 import httpx
-from pydantic import BaseModel, ConfigDict, TypeAdapter
+from pydantic import ConfigDict, TypeAdapter
 
 from litellm.litellm_core_utils.core_helpers import map_finish_reason
 from litellm.llms.base_llm.chat.transformation import BaseConfig, BaseLLMException
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import AllMessageValues
 from litellm.types.utils import Message, ModelResponse, Usage
 
@@ -30,14 +31,14 @@ REASONING_DISABLED_EFFORTS: Final[frozenset[str]] = frozenset(("none", "minimal"
 REASONING_ENABLED_EFFORTS: Final[frozenset[str]] = frozenset(("low", "medium", "high"))
 
 
-class _FalUsage(BaseModel):
+class _FalUsage(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     input_tokens: int
     output_tokens: int
 
 
-class _FalChatResponse(BaseModel):
+class _FalChatResponse(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     output: str
diff --git a/litellm/llms/fireworks_ai/rerank/transformation.py b/litellm/llms/fireworks_ai/rerank/transformation.py
index 3d9d813c6e9..4eff64e351f 100644
--- a/litellm/llms/fireworks_ai/rerank/transformation.py
+++ b/litellm/llms/fireworks_ai/rerank/transformation.py
@@ -8,12 +8,13 @@ from collections.abc import Iterable, Mapping, Sequence
 from typing import Final
 
 import httpx
-from pydantic import BaseModel, ConfigDict, TypeAdapter
+from pydantic import ConfigDict, TypeAdapter
 
 from litellm._uuid import uuid
 from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
 from litellm.llms.base_llm.rerank.transformation import BaseRerankConfig
 from litellm.llms.fireworks_ai.common_utils import FireworksAIMixin
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.rerank import (
     RerankBilledUnits,
     RerankResponse,
@@ -24,7 +25,7 @@ from litellm.types.rerank import (
 )
 
 
-class _FireworksAIUsageFields(BaseModel):
+class _FireworksAIUsageFields(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     total_tokens: int | None = 0
@@ -32,7 +33,7 @@ class _FireworksAIUsageFields(BaseModel):
     completion_tokens: int | None = 0
 
 
-class _FireworksAIResultFields(BaseModel):
+class _FireworksAIResultFields(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True, hide_input_in_errors=True)
 
     index: int | float | str
diff --git a/litellm/llms/groq/chat/transformation.py b/litellm/llms/groq/chat/transformation.py
index f60947714a5..cb3f1ac2d0c 100644
--- a/litellm/llms/groq/chat/transformation.py
+++ b/litellm/llms/groq/chat/transformation.py
@@ -16,6 +16,7 @@ from litellm.llms.openai.chat.gpt_transformation import (
 )
 from litellm.llms.openai.common_utils import OpenAIError
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import (
     AllMessageValues,
     ChatCompletionAssistantMessage,
@@ -32,7 +33,7 @@ if TYPE_CHECKING:
 GROQ_COMPOUND_MODELS: Final = frozenset({"compound", "compound-mini"})
 
 
-class GroqExecutedToolIdentity(BaseModel):
+class GroqExecutedToolIdentity(LiteLLMBaseModel):
     name: str | None = None
     type: str | None = None
 
diff --git a/litellm/llms/laya/common_utils.py b/litellm/llms/laya/common_utils.py
index 3e423a9e742..ff6343b06db 100644
--- a/litellm/llms/laya/common_utils.py
+++ b/litellm/llms/laya/common_utils.py
@@ -1,10 +1,12 @@
 from collections.abc import Mapping
 from typing import Final
 
-from pydantic import BaseModel, TypeAdapter, ValidationError
+from pydantic import TypeAdapter, ValidationError
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class _LayaRouting(BaseModel):
+class _LayaRouting(LiteLLMBaseModel):
     model: str | None = None
 
 
diff --git a/litellm/llms/litellm_proxy/skills/skill_search.py b/litellm/llms/litellm_proxy/skills/skill_search.py
index f975c6c4cab..adfd45b7d11 100644
--- a/litellm/llms/litellm_proxy/skills/skill_search.py
+++ b/litellm/llms/litellm_proxy/skills/skill_search.py
@@ -6,7 +6,7 @@ from collections.abc import Sequence
 from dataclasses import dataclass
 from typing import TYPE_CHECKING, Final, TypeAlias
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 
 from litellm.llms.litellm_proxy.skills.constants import MAX_SKILLS_PER_SEARCH
 from litellm.llms.litellm_proxy.skills.handler import LiteLLMSkillsHandler
@@ -16,6 +16,7 @@ from litellm.proxy.common_utils.semantic_text_index import (
     SemanticTextIndex,
     router_embedder,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import LlmProviders
 
 if TYPE_CHECKING:
@@ -63,7 +64,7 @@ SkillSearchOutcome: TypeAlias = SkillSearchHits | SkillSearchNotConfigured | Ski
 HostedSkillSearchOutcome: TypeAlias = SkillSearchOutcome | SkillSearchUnsupportedProvider
 
 
-class SkillSearchResult(BaseModel):
+class SkillSearchResult(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     skill_id: str
diff --git a/litellm/llms/mistral/batches/transformation.py b/litellm/llms/mistral/batches/transformation.py
index 3496feed585..86de2836048 100644
--- a/litellm/llms/mistral/batches/transformation.py
+++ b/litellm/llms/mistral/batches/transformation.py
@@ -15,12 +15,13 @@ import httpx
 from openai.types.batch import BatchRequestCounts
 from openai.types.batch import Errors as BatchErrors
 from openai.types.batch_error import BatchError
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 from typing_extensions import NotRequired, ReadOnly, TypedDict
 
 from litellm.litellm_core_utils.url_utils import encode_url_path_segment
 from litellm.llms.base_llm.batches.transformation import BaseBatchesConfig
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import AllMessageValues, CreateBatchRequest
 from litellm.types.utils import LiteLLMBatch, LlmProviders
 
@@ -64,14 +65,14 @@ class MistralPresignedRequest(TypedDict):
     headers: ReadOnly[Mapping[str, str]]
 
 
-class MistralBatchError(BaseModel):
+class MistralBatchError(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     message: str
     count: int = 1
 
 
-class MistralBatchJob(BaseModel):
+class MistralBatchJob(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     id: str
diff --git a/litellm/llms/mistral/files/transformation.py b/litellm/llms/mistral/files/transformation.py
index c1e3f50c379..86bce763917 100644
--- a/litellm/llms/mistral/files/transformation.py
+++ b/litellm/llms/mistral/files/transformation.py
@@ -14,13 +14,14 @@ from typing import Final, Literal, TypeAlias
 
 import httpx
 from openai.types.file_deleted import FileDeleted
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 from typing_extensions import ReadOnly, TypedDict
 
 from litellm.litellm_core_utils.prompt_templates.common_utils import extract_file_data
 from litellm.litellm_core_utils.url_utils import encode_url_path_segment
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
 from litellm.llms.base_llm.files.transformation import BaseFilesConfig, LiteLLMLoggingObj
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import (
     CreateFileRequest,
     FileContentRequest,
@@ -54,7 +55,7 @@ class MistralMultipartUpload(TypedDict):
     purpose: ReadOnly[tuple[None, MistralFilePurpose]]
 
 
-class MistralFile(BaseModel):
+class MistralFile(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     id: str
@@ -65,13 +66,13 @@ class MistralFile(BaseModel):
     expires_at: int | None = None
 
 
-class MistralFileList(BaseModel):
+class MistralFileList(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     data: tuple[MistralFile, ...] = ()
 
 
-class MistralFileDeleted(BaseModel):
+class MistralFileDeleted(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     id: str
diff --git a/litellm/llms/mongodb/vector_stores/transformation.py b/litellm/llms/mongodb/vector_stores/transformation.py
index c6d3a2db3b4..0d931ee78cb 100644
--- a/litellm/llms/mongodb/vector_stores/transformation.py
+++ b/litellm/llms/mongodb/vector_stores/transformation.py
@@ -6,7 +6,7 @@ from typing import TYPE_CHECKING, Final, Literal, NoReturn
 from urllib.parse import quote, urlsplit
 
 import httpx
-from pydantic import BaseModel, ConfigDict, TypeAdapter, ValidationError
+from pydantic import ConfigDict, TypeAdapter, ValidationError
 
 from litellm.exceptions import AuthenticationError, BadRequestError, ServiceUnavailableError, Timeout
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
@@ -16,6 +16,7 @@ from litellm.llms.base_llm.vector_store.transformation import (
     VectorStoreEmbeddingExecutor,
 )
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.router import GenericLiteLLMParams
 from litellm.types.utils import EmbeddingResponse
 from litellm.types.vector_stores import (
@@ -49,13 +50,13 @@ def config_error(message: str) -> BadRequestError:
     return BadRequestError(message=message, model=None, llm_provider="mongodb")
 
 
-class _Content(BaseModel):
+class _Content(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, strict=True)
     type: Literal["text"]
     text: str
 
 
-class _Result(BaseModel):
+class _Result(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, strict=True, allow_inf_nan=False)
     score: float | None
     content: Sequence[_Content]
@@ -63,14 +64,14 @@ class _Result(BaseModel):
     filename: str | None
 
 
-class _SearchResponse(BaseModel):
+class _SearchResponse(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, strict=True)
     object: Literal["vector_store.search_results.page"]
     search_query: str
     data: Sequence[_Result]
 
 
-class _MongoDBSearchParams(BaseModel):
+class _MongoDBSearchParams(LiteLLMBaseModel):
     """Typed view over the vector store's litellm_params; unrelated keys are ignored."""
 
     model_config = ConfigDict(frozen=True, extra="ignore")
diff --git a/litellm/llms/nimble/search/transformation.py b/litellm/llms/nimble/search/transformation.py
index f40ca60fb6c..30cfc38eb87 100644
--- a/litellm/llms/nimble/search/transformation.py
+++ b/litellm/llms/nimble/search/transformation.py
@@ -11,7 +11,7 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Final
 
 import httpx
-from pydantic import BaseModel, ConfigDict, TypeAdapter, ValidationError
+from pydantic import ConfigDict, TypeAdapter, ValidationError
 
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
 from litellm.llms.base_llm.search.transformation import (
@@ -20,6 +20,7 @@ from litellm.llms.base_llm.search.transformation import (
     SearchResult,
 )
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
@@ -27,7 +28,7 @@ if TYPE_CHECKING:
 _NIMBLE_DOCS_URL: Final = "https://docs.nimbleway.com/api-reference/search/search"
 
 
-class _NimbleResult(BaseModel):
+class _NimbleResult(LiteLLMBaseModel):
     """One entry of Nimble's `results` array. Every field is optional so a single degraded
     result degrades to empty strings instead of failing the whole call."""
 
@@ -41,7 +42,7 @@ class _NimbleResult(BaseModel):
     additional_data: object = None
 
 
-class _NimbleSearchResponse(BaseModel):
+class _NimbleSearchResponse(LiteLLMBaseModel):
     """Nimble's /v2/search response envelope."""
 
     model_config = ConfigDict(extra="ignore", frozen=True)
@@ -51,7 +52,7 @@ class _NimbleSearchResponse(BaseModel):
     results: tuple[_NimbleResult, ...]
 
 
-class _AdditionalData(BaseModel):
+class _AdditionalData(LiteLLMBaseModel):
     """The slice of a result's free-form `additional_data` that maps onto SearchResult."""
 
     model_config = ConfigDict(extra="ignore", frozen=True)
@@ -59,7 +60,7 @@ class _AdditionalData(BaseModel):
     publish_date: str | None = None
 
 
-class _ErrorEnvelope(BaseModel):
+class _ErrorEnvelope(LiteLLMBaseModel):
     """Nimble reports errors as either `{"detail": ...}` (validation) or
     `{"success": "false", "task_id": ..., "message": ...}` (collection)."""
 
diff --git a/litellm/llms/oci/common_utils.py b/litellm/llms/oci/common_utils.py
index 679a9c21e43..fe8d0e1c1e7 100644
--- a/litellm/llms/oci/common_utils.py
+++ b/litellm/llms/oci/common_utils.py
@@ -13,10 +13,11 @@ from typing import Final, Protocol, runtime_checkable
 from urllib.parse import urlparse
 
 import httpx
-from pydantic import BaseModel, ConfigDict, Field, JsonValue, TypeAdapter, ValidationError, field_validator
+from pydantic import ConfigDict, Field, JsonValue, TypeAdapter, ValidationError, field_validator
 
 from litellm._logging import verbose_logger
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
+from litellm.types.llms.base import LiteLLMBaseModel
 
 try:
     from cryptography.hazmat.primitives import hashes, serialization
@@ -214,7 +215,7 @@ _OCI_REALM_DOMAINS: Final = MappingProxyType(
 )
 
 
-class OCIRegionMetadata(BaseModel):
+class OCIRegionMetadata(LiteLLMBaseModel):
     """One entry of the OCI SDK's region metadata schema, as found in
     ``~/.oci/regions-config.json`` (a JSON array) or ``OCI_REGION_METADATA`` (one object).
     Values are lowercased before validation, as the SDK does."""
diff --git a/litellm/llms/ollama/completion/transformation.py b/litellm/llms/ollama/completion/transformation.py
index b1f69220de7..9687378e5c3 100644
--- a/litellm/llms/ollama/completion/transformation.py
+++ b/litellm/llms/ollama/completion/transformation.py
@@ -4,7 +4,7 @@ from collections.abc import AsyncIterator, Iterator
 from typing import TYPE_CHECKING, Any, Final
 
 from httpx._models import Headers, Response
-from pydantic import BaseModel, ConfigDict, ValidationError
+from pydantic import ConfigDict, ValidationError
 
 import litellm
 from litellm._logging import verbose_proxy_logger
@@ -23,6 +23,7 @@ from litellm.litellm_core_utils.prompt_templates.image_handling import (
 )
 from litellm.llms.base_llm.base_model_iterator import BaseModelResponseIterator
 from litellm.llms.base_llm.chat.transformation import BaseConfig, BaseLLMException
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import AllMessageValues, ChatCompletionUsageBlock
 from litellm.types.utils import (
     Delta,
@@ -44,7 +45,7 @@ else:
     LiteLLMLoggingObj = Any
 
 
-class _OllamaGenerateReasoning(BaseModel):
+class _OllamaGenerateReasoning(LiteLLMBaseModel):
     """The two `/api/generate` fields a reply's reasoning can arrive in."""
 
     model_config = ConfigDict(extra="ignore")
diff --git a/litellm/llms/openai/organization_costs.py b/litellm/llms/openai/organization_costs.py
index 8e7f02cca96..4f1d4798e99 100644
--- a/litellm/llms/openai/organization_costs.py
+++ b/litellm/llms/openai/organization_costs.py
@@ -8,7 +8,7 @@ from types import MappingProxyType
 from typing import Final, Literal, TypeAlias
 
 import httpx
-from pydantic import BaseModel, ConfigDict, ValidationError
+from pydantic import ConfigDict, ValidationError
 
 from litellm.constants import (
     OPENAI_ORGANIZATION_COSTS_PAGE_LIMIT,
@@ -16,6 +16,7 @@ from litellm.constants import (
     PROVIDER_BILLING_TIMEOUT_SECONDS,
 )
 from litellm.llms.custom_httpx.http_handler import get_async_httpx_client
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.custom_http import httpxSpecialProvider
 
 OPENAI_ADMIN_KEY_ENV_VAR: Final = "OPENAI_ADMIN_KEY"
@@ -31,27 +32,27 @@ class OpenAICostsRequestFailed:
     detail: str
 
 
-class _OpenAICostAmount(BaseModel):
+class _OpenAICostAmount(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     value: float
     currency: Literal["usd"]
 
 
-class _OpenAICostResult(BaseModel):
+class _OpenAICostResult(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     amount: _OpenAICostAmount
 
 
-class _OpenAICostBucket(BaseModel):
+class _OpenAICostBucket(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     start_time: int
     results: tuple[_OpenAICostResult, ...] = ()
 
 
-class _OpenAICostsPage(BaseModel):
+class _OpenAICostsPage(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     data: tuple[_OpenAICostBucket, ...]
diff --git a/litellm/llms/openai/responses/guardrail_translation/handler.py b/litellm/llms/openai/responses/guardrail_translation/handler.py
index ad68924ce20..a267b66bb22 100644
--- a/litellm/llms/openai/responses/guardrail_translation/handler.py
+++ b/litellm/llms/openai/responses/guardrail_translation/handler.py
@@ -66,6 +66,7 @@ from litellm.llms.openai.responses.guardrail_translation.tool_merge import merge
 from litellm.responses.litellm_completion_transformation.transformation import (
     LiteLLMCompletionResponsesConfig,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import (
     AllMessageValues,
     BaseLiteLLMOpenAIResponseObject,
@@ -110,14 +111,14 @@ class _ToolCallShape(NamedTuple):
     arguments: str
 
 
-class _ToolCallFunctionFields(BaseModel):
+class _ToolCallFunctionFields(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     name: str | None = None
     arguments: str = ""
 
 
-class _ToolCallFields(BaseModel):
+class _ToolCallFields(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     function: _ToolCallFunctionFields
diff --git a/litellm/llms/openai/responses/transformation.py b/litellm/llms/openai/responses/transformation.py
index 1452ecebca9..e4928868699 100644
--- a/litellm/llms/openai/responses/transformation.py
+++ b/litellm/llms/openai/responses/transformation.py
@@ -26,6 +26,7 @@ from litellm.llms.openai.chat.gpt_5_transformation import is_gpt_reasoning_serie
 from litellm.responses.litellm_completion_transformation.custom_tools import TOOL_CALL_ITEM_ID_PREFIX_BY_TYPE
 from litellm.responses.litellm_completion_transformation.reasoning_items import is_litellm_minted_reasoning_item
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import *
 from litellm.types.responses.main import *
 from litellm.types.router import GenericLiteLLMParams
@@ -51,7 +52,7 @@ _PROVIDERS_VALIDATING_TOOL_CALL_ITEM_IDS: Final = frozenset({LlmProviders.AZURE,
 _PROVIDERS_REPLAYING_ONLY_THEIR_OWN_REASONING: Final = _PROVIDERS_VALIDATING_TOOL_CALL_ITEM_IDS
 
 
-class _ReasoningSupportEntry(BaseModel):
+class _ReasoningSupportEntry(LiteLLMBaseModel):
     litellm_provider: str | None = None
     supports_reasoning: bool | None = None
 
diff --git a/litellm/llms/openai_like/model_info.py b/litellm/llms/openai_like/model_info.py
index 101be58d197..24135287ae9 100644
--- a/litellm/llms/openai_like/model_info.py
+++ b/litellm/llms/openai_like/model_info.py
@@ -5,11 +5,12 @@ from types import MappingProxyType
 from typing import Annotated, Final, TypeAlias
 
 import httpx
-from pydantic import BaseModel, BeforeValidator, ConfigDict
+from pydantic import BeforeValidator, ConfigDict
 
 from litellm._logging import verbose_logger
 from litellm.caching.in_memory_cache import InMemoryCache
 from litellm.llms.custom_httpx.http_handler import AsyncHTTPHandler
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.utils import _add_path_to_api_base  # pyright: ignore[reportPrivateUsage]  # shared provider URL helper
 
 MODEL_INFO_REFRESH_SECONDS: Final = 300
@@ -25,7 +26,7 @@ def _positive_limit(value: object) -> int | None:
 _TokenLimit: TypeAlias = Annotated[int | None, BeforeValidator(_positive_limit)]
 
 
-class _ModelCard(BaseModel):
+class _ModelCard(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     id: str
@@ -51,7 +52,7 @@ class _ModelCard(BaseModel):
         )
 
 
-class _ModelList(BaseModel):
+class _ModelList(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     data: tuple[_ModelCard, ...] = ()
diff --git a/litellm/llms/parallel_ai/search/transformation.py b/litellm/llms/parallel_ai/search/transformation.py
index d91e532a2cf..c627cb001dc 100644
--- a/litellm/llms/parallel_ai/search/transformation.py
+++ b/litellm/llms/parallel_ai/search/transformation.py
@@ -9,7 +9,7 @@ from types import MappingProxyType
 from typing import Final, TypedDict
 
 import httpx
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 from typing_extensions import ReadOnly
 
 from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
@@ -20,9 +20,10 @@ from litellm.llms.base_llm.search.transformation import (
 )
 from litellm.llms.parallel_ai.search.cost_calculator import PARALLEL_AI_USAGE_PARAM
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class _ParallelAIV1SearchResult(BaseModel):
+class _ParallelAIV1SearchResult(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     url: str | None = None
@@ -31,7 +32,7 @@ class _ParallelAIV1SearchResult(BaseModel):
     excerpts: Sequence[str] | None = None
 
 
-class _ParallelAIV1SearchResponse(BaseModel):
+class _ParallelAIV1SearchResponse(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     search_id: str | None = None
diff --git a/litellm/llms/sap/chat/models.py b/litellm/llms/sap/chat/models.py
index 5f65c7f715d..afdd7a2f356 100644
--- a/litellm/llms/sap/chat/models.py
+++ b/litellm/llms/sap/chat/models.py
@@ -2,7 +2,9 @@ import warnings
 from enum import Enum
 from typing import Final, Literal
 
-from pydantic import BaseModel, Field, field_validator, model_validator
+from pydantic import Field, field_validator, model_validator
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
 def validate_different_content(v: str | dict | list) -> str:
@@ -24,30 +26,30 @@ def validate_different_content(v: str | dict | list) -> str:
     raise ValueError("Content must be a string")
 
 
-class TextContent(BaseModel):
+class TextContent(LiteLLMBaseModel):
     type_: Literal["text"] = Field(default="text", alias="type")
     text: str
 
 
-class ImageURLContent(BaseModel):
+class ImageURLContent(LiteLLMBaseModel):
     url: str
     detail: str = "auto"
 
 
-class ImageContent(BaseModel):
+class ImageContent(LiteLLMBaseModel):
     type_: Literal["image_url"] = Field(default="image_url", alias="type")
     image_url: ImageURLContent
 
 
-class FunctionObj(BaseModel):
+class FunctionObj(LiteLLMBaseModel):
     name: str
     arguments: str
 
 
-class FunctionTool(BaseModel):
+class FunctionTool(LiteLLMBaseModel):
     description: str = ""
     name: str
-    parameters: dict = {"type": "object", "properties": {}}
+    parameters: dict = Field(default={"type": "object", "properties": {}})
     strict: bool = False
 
     def model_dump(self, **kwargs) -> dict:
@@ -67,7 +69,7 @@ class FunctionTool(BaseModel):
         return v
 
 
-class ChatCompletionTool(BaseModel):
+class ChatCompletionTool(LiteLLMBaseModel):
     type_: Literal["function"] = Field(default="function", alias="type")
     function: FunctionTool
 
@@ -76,13 +78,13 @@ class ChatCompletionTool(BaseModel):
         return super().model_dump(**kwargs)
 
 
-class MessageToolCall(BaseModel):
+class MessageToolCall(LiteLLMBaseModel):
     id: str
     type_: Literal["function"] = Field(default="function", alias="type")
     function: FunctionObj
 
 
-class SAPMessage(BaseModel):
+class SAPMessage(LiteLLMBaseModel):
     """
     Model for SystemChatMessage and DeveloperChatMessage
     """
@@ -93,21 +95,21 @@ class SAPMessage(BaseModel):
     _content_validator = field_validator("content", mode="before")(validate_different_content)
 
 
-class SAPUserMessage(BaseModel):
+class SAPUserMessage(LiteLLMBaseModel):
     role: Literal["user"] = "user"
     content: str | TextContent | ImageContent | list[TextContent | ImageContent]
 
 
-class SAPAssistantMessage(BaseModel):
+class SAPAssistantMessage(LiteLLMBaseModel):
     role: Literal["assistant"] = "assistant"
     content: str = ""
     refusal: str = ""
-    tool_calls: list[MessageToolCall] = []
+    tool_calls: list[MessageToolCall] = Field(default=[])
 
     _content_validator = field_validator("content", mode="before")(validate_different_content)
 
 
-class SAPToolChatMessage(BaseModel):
+class SAPToolChatMessage(LiteLLMBaseModel):
     role: Literal["tool"] = "tool"
     tool_call_id: str
     content: str
@@ -118,23 +120,23 @@ class SAPToolChatMessage(BaseModel):
 ChatMessage = SAPMessage | SAPUserMessage | SAPAssistantMessage | SAPToolChatMessage
 
 
-class ResponseFormat(BaseModel):
+class ResponseFormat(LiteLLMBaseModel):
     type_: Literal["text", "json_object"] = Field(default="text", alias="type")
 
 
-class JSONResponseSchema(BaseModel):
+class JSONResponseSchema(LiteLLMBaseModel):
     description: str = ""
     name: str
     schema_: dict = Field(default_factory=dict, alias="schema")
     strict: bool = False
 
 
-class ResponseFormatJSONSchema(BaseModel):
+class ResponseFormatJSONSchema(LiteLLMBaseModel):
     type_: Literal["json_schema"] = Field(default="json_schema", alias="type")
     json_schema: JSONResponseSchema
 
 
-class KeyValueListPair(BaseModel):
+class KeyValueListPair(LiteLLMBaseModel):
     key: str
     value: list[str]
 
@@ -143,7 +145,7 @@ class DocumentMetadataKeyValueListPairs(KeyValueListPair):
     select_mode: list[Literal["ignoreIfKeyAbsent"]] | None = None
 
 
-class GroundingSearchConfig(BaseModel):
+class GroundingSearchConfig(LiteLLMBaseModel):
     max_chunk_count: int | None = Field(default=None, ge=0)
     max_document_count: int | None = Field(default=None, ge=0)
 
@@ -154,7 +156,7 @@ class GroundingSearchConfig(BaseModel):
         return self
 
 
-class DocumentGroundingFilter(BaseModel):
+class DocumentGroundingFilter(LiteLLMBaseModel):
     id_: str | None = Field(default=None, alias="id")
     data_repository_type: Literal["vector", "help.sap.com"]
     search_config: GroundingSearchConfig | None = None
@@ -164,36 +166,36 @@ class DocumentGroundingFilter(BaseModel):
     chunk_metadata: list[KeyValueListPair] | None = None
 
 
-class DocumentGroundingPlaceholders(BaseModel):
+class DocumentGroundingPlaceholders(LiteLLMBaseModel):
     input: list[str] = Field(min_length=1)
     output: str
 
 
-class DocumentGroundingConfig(BaseModel):
+class DocumentGroundingConfig(LiteLLMBaseModel):
     filters: list[DocumentGroundingFilter] | None = None
     placeholders: DocumentGroundingPlaceholders
     metadata_params: list[str] | None = None
 
 
-class GroundingModuleConfig(BaseModel):
+class GroundingModuleConfig(LiteLLMBaseModel):
     type_: Literal["document_grounding_service"] = Field(default="document_grounding_service", alias="type")
     config: DocumentGroundingConfig
 
 
-class Template(BaseModel):
+class Template(LiteLLMBaseModel):
     template: list[ChatMessage]
     defaults: dict[str, str] | None = None
     response_format: ResponseFormat | ResponseFormatJSONSchema | None = None
     tools: list[ChatCompletionTool] | None = None
 
 
-class LLMModelDetails(BaseModel):
+class LLMModelDetails(LiteLLMBaseModel):
     name: str
     version: str = "latest"
     params: dict | None = None
 
 
-class PromptTemplatingModuleConfig(BaseModel):
+class PromptTemplatingModuleConfig(LiteLLMBaseModel):
     prompt: Template
     model: LLMModelDetails
 
@@ -285,7 +287,7 @@ class SAPMaskingProfileEntity(str, Enum):
     ETHNICITY = "profile-ethnicity"
 
 
-class DPIMethodConstant(BaseModel):
+class DPIMethodConstant(LiteLLMBaseModel):
     """
     Replaces the entity with the specified value followed by an incrementing number
     """
@@ -294,7 +296,7 @@ class DPIMethodConstant(BaseModel):
     value: str
 
 
-class DPIMethodFabricatedData(BaseModel):
+class DPIMethodFabricatedData(LiteLLMBaseModel):
     """
     Replaces the entity with a randomly generated value appropriate to its type.
     """
@@ -302,7 +304,7 @@ class DPIMethodFabricatedData(BaseModel):
     method: Literal["fabricated_data"] = "fabricated_data"
 
 
-class DPICustomEntity(BaseModel):
+class DPICustomEntity(LiteLLMBaseModel):
     """
     regex: Regular expression to match the entity
     replacement_strategy: Replacement strategy to be used for the entity
@@ -312,7 +314,7 @@ class DPICustomEntity(BaseModel):
     replacement_strategy: DPIMethodConstant
 
 
-class DPIStandardEntity(BaseModel):
+class DPIStandardEntity(LiteLLMBaseModel):
     """
     type: Standard entity type to be masked
     replacement_strategy: Replacement strategy to be used for the entity
@@ -322,7 +324,7 @@ class DPIStandardEntity(BaseModel):
     replacement_strategy: DPIMethodConstant | DPIMethodFabricatedData | None = None
 
 
-class MaskGroundingInput(BaseModel):
+class MaskGroundingInput(LiteLLMBaseModel):
     """
     Controls whether the input to the grounding module will be masked with the configuration
     supplied in the masking module
@@ -331,7 +333,7 @@ class MaskGroundingInput(BaseModel):
     enabled: bool = False
 
 
-class MaskingProviderConfig(BaseModel):
+class MaskingProviderConfig(LiteLLMBaseModel):
     """
     SAP Data Privacy Integration provider for data masking.
 
@@ -356,7 +358,7 @@ class MaskingProviderConfig(BaseModel):
     mask_grounding_input: MaskGroundingInput | None = None
 
 
-class MaskingModuleConfig(BaseModel):
+class MaskingModuleConfig(LiteLLMBaseModel):
     """
     Configuration for the data masking module.
 
@@ -417,7 +419,7 @@ class AzureThreshold(int, Enum):
     ALLOW_ALL = 6
 
 
-class AzureContentFilter(BaseModel):
+class AzureContentFilter(LiteLLMBaseModel):
     """
     Specific filter configuration for Azure Content Safety.
 
@@ -481,7 +483,7 @@ class AzureContentSafetyOutput(AzureContentFilter):
     protected_material_code: bool | None = False
 
 
-class LlamaGuard38bFilter(BaseModel):
+class LlamaGuard38bFilter(LiteLLMBaseModel):
     """
     Specific implementation of ContentFilter for Llama Guard 3. Llama Guard 3 is a
     Llama-3.1-8B pretrained model, fine-tuned for content safety classification.
@@ -532,22 +534,22 @@ class LlamaGuard38bFilter(BaseModel):
     code_interpreter_abuse: bool = Field(default=False)
 
 
-class LlamaGuard38bFilterConfig(BaseModel):
+class LlamaGuard38bFilterConfig(LiteLLMBaseModel):
     type_: Literal["llama_guard_3_8b"] = Field(default="llama_guard_3_8b", alias="type")
     config: LlamaGuard38bFilter
 
 
-class AzureContentSafetyInputFilterConfig(BaseModel):
+class AzureContentSafetyInputFilterConfig(LiteLLMBaseModel):
     type_: Literal["azure_content_safety"] = Field(default="azure_content_safety", alias="type")
     config: AzureContentSafetyInput | None = None
 
 
-class AzureContentSafetyOutputFilterConfig(BaseModel):
+class AzureContentSafetyOutputFilterConfig(LiteLLMBaseModel):
     type_: Literal["azure_content_safety"] = Field(default="azure_content_safety", alias="type")
     config: AzureContentSafetyOutput | None = None
 
 
-class FilteringStreamOptions(BaseModel):
+class FilteringStreamOptions(LiteLLMBaseModel):
     """
     overlap: Number of characters that should be additionally sent to content filtering services
     from previous chunks as additional context.
@@ -556,7 +558,7 @@ class FilteringStreamOptions(BaseModel):
     overlap: int | None = Field(default=0, ge=0, le=10000)
 
 
-class InputFiltering(BaseModel):
+class InputFiltering(LiteLLMBaseModel):
     """Module for managing and applying input content filters.
 
     Args:
@@ -566,7 +568,7 @@ class InputFiltering(BaseModel):
     filters: list[AzureContentSafetyInputFilterConfig | LlamaGuard38bFilterConfig] = Field(min_length=1)
 
 
-class OutputFiltering(BaseModel):
+class OutputFiltering(LiteLLMBaseModel):
     """Module for managing and applying output content filters.
 
     Args:
@@ -579,7 +581,7 @@ class OutputFiltering(BaseModel):
     stream_options: FilteringStreamOptions | None = None
 
 
-class FilteringModuleConfig(BaseModel):
+class FilteringModuleConfig(LiteLLMBaseModel):
     """Module for managing and applying content filters.
 
     Args:
@@ -603,7 +605,7 @@ class FilteringModuleConfig(BaseModel):
         return self
 
 
-class SAPDocumentTranslationApplyToSelector(BaseModel):
+class SAPDocumentTranslationApplyToSelector(LiteLLMBaseModel):
     """
     This selector allows you to define the scope of translation, such as specific placeholders or
     messages with specific roles.
@@ -619,7 +621,7 @@ class SAPDocumentTranslationApplyToSelector(BaseModel):
     source_language: str
 
 
-class InputTranslationConfig(BaseModel):
+class InputTranslationConfig(LiteLLMBaseModel):
     """
     Configuration for input translation.
 
@@ -634,12 +636,12 @@ class InputTranslationConfig(BaseModel):
     apply_to: list[SAPDocumentTranslationApplyToSelector] | None = None
 
 
-class OutputTranslationConfig(BaseModel):
+class OutputTranslationConfig(LiteLLMBaseModel):
     source_language: str | None = None
     target_language: str | SAPDocumentTranslationApplyToSelector
 
 
-class SAPDocumentTranslationInput(BaseModel):
+class SAPDocumentTranslationInput(LiteLLMBaseModel):
     """
     Configuration for input translation
 
@@ -656,7 +658,7 @@ class SAPDocumentTranslationInput(BaseModel):
     config: InputTranslationConfig
 
 
-class SAPDocumentTranslationOutput(BaseModel):
+class SAPDocumentTranslationOutput(LiteLLMBaseModel):
     """
     Configuration for output translation
 
@@ -670,7 +672,7 @@ class SAPDocumentTranslationOutput(BaseModel):
     config: OutputTranslationConfig
 
 
-class TranslationModuleConfig(BaseModel):
+class TranslationModuleConfig(LiteLLMBaseModel):
     """
     Configuration for translation module
 
@@ -690,7 +692,7 @@ class TranslationModuleConfig(BaseModel):
         return self
 
 
-class ModuleConfig(BaseModel):
+class ModuleConfig(LiteLLMBaseModel):
     prompt_templating: PromptTemplatingModuleConfig
     filtering: FilteringModuleConfig | None = None
     masking: MaskingModuleConfig | None = None
@@ -698,17 +700,17 @@ class ModuleConfig(BaseModel):
     translation: TranslationModuleConfig | None = None
 
 
-class GlobalStreamOptions(BaseModel):
+class GlobalStreamOptions(LiteLLMBaseModel):
     enabled: bool = False
     chunk_size: int | None = Field(default=None, ge=1)
     delimiters: list[str] | None = None
 
 
-class OrchestrationConfig(BaseModel):
+class OrchestrationConfig(LiteLLMBaseModel):
     modules: ModuleConfig | list[ModuleConfig]
     stream: GlobalStreamOptions | None = None
 
 
-class OrchestrationRequest(BaseModel):
+class OrchestrationRequest(LiteLLMBaseModel):
     config: OrchestrationConfig
     placeholder_values: dict[str, str] | None = None
diff --git a/litellm/llms/sap/embed/transformation.py b/litellm/llms/sap/embed/transformation.py
index 4a3062c8cbc..e91dfb91081 100644
--- a/litellm/llms/sap/embed/transformation.py
+++ b/litellm/llms/sap/embed/transformation.py
@@ -6,13 +6,14 @@ from functools import cached_property
 from typing import Final, Literal
 
 import httpx
-from pydantic import BaseModel, Field
+from pydantic import Field
 
 from litellm.llms.base_llm.embedding.transformation import (
     BaseEmbeddingConfig,
     LiteLLMLoggingObj,
 )
 from litellm.llms.sap.chat.models import MaskingModuleConfig
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import AllEmbeddingInputValues
 from litellm.types.utils import EmbeddingResponse
 
@@ -20,30 +21,30 @@ from ..chat.handler import GenAIHubOrchestrationError
 from ..credentials import get_token_creator
 
 
-class Usage(BaseModel):
+class Usage(LiteLLMBaseModel):
     prompt_tokens: int
     total_tokens: int
 
 
-class EmbeddingItem(BaseModel):
+class EmbeddingItem(LiteLLMBaseModel):
     object: Literal["embedding"]
     embedding: list[float] = Field(..., description="Vector of floats (length varies by model).")
     index: int
 
 
-class FinalResult(BaseModel):
+class FinalResult(LiteLLMBaseModel):
     object: Literal["list"]
     data: list[EmbeddingItem]
     model: str
     usage: Usage
 
 
-class EmbeddingsResponse(BaseModel):
+class EmbeddingsResponse(LiteLLMBaseModel):
     request_id: str
     final_result: FinalResult
 
 
-class EmbeddingModel(BaseModel):
+class EmbeddingModel(LiteLLMBaseModel):
     name: str
     version: str = "latest"
     params: dict = Field(default_factory=dict)
@@ -51,25 +52,25 @@ class EmbeddingModel(BaseModel):
     max_retries: int | None = Field(default=None, ge=0, le=5)
 
 
-class EmbeddingsModelConfig(BaseModel):
+class EmbeddingsModelConfig(LiteLLMBaseModel):
     model: EmbeddingModel
 
 
-class EmbeddingsModules(BaseModel):
+class EmbeddingsModules(LiteLLMBaseModel):
     embeddings: EmbeddingsModelConfig
     masking: MaskingModuleConfig | None = None
 
 
-class EmbeddingInput(BaseModel):
+class EmbeddingInput(LiteLLMBaseModel):
     text: str | list[str]
     type: Literal["text", "document", "query"] | None = None
 
 
-class EmbeddingConfig(BaseModel):
+class EmbeddingConfig(LiteLLMBaseModel):
     modules: EmbeddingsModules
 
 
-class EmbeddingRequest(BaseModel):
+class EmbeddingRequest(LiteLLMBaseModel):
     config: EmbeddingConfig
     input: EmbeddingInput
 
diff --git a/litellm/llms/valkey/vector_stores/transformation.py b/litellm/llms/valkey/vector_stores/transformation.py
index 50485899818..cd95d9c95b7 100644
--- a/litellm/llms/valkey/vector_stores/transformation.py
+++ b/litellm/llms/valkey/vector_stores/transformation.py
@@ -12,7 +12,7 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Final, NoReturn
 
 import httpx
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 
 import litellm
 from litellm.llms.base_llm.vector_store.transformation import (
@@ -20,6 +20,7 @@ from litellm.llms.base_llm.vector_store.transformation import (
     VectorStoreEmbeddingExecutor,
 )
 from litellm.llms.valkey.common_utils import build_valkey_url, pack_vector
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import EmbeddingResponse
 from litellm.types.vector_stores import (
     VectorStoreCreateOptionalRequestParams,
@@ -79,7 +80,7 @@ def _import_query() -> "type[Query]":
     return RedisQuery
 
 
-class _ValkeySearchParams(BaseModel):
+class _ValkeySearchParams(LiteLLMBaseModel):
     """Typed view over the vector store's litellm_params; unrelated keys are ignored."""
 
     model_config = ConfigDict(frozen=True, extra="ignore")
diff --git a/litellm/llms/vertex_ai/vertex_embeddings/transformation.py b/litellm/llms/vertex_ai/vertex_embeddings/transformation.py
index e4bbdd1bd0d..cab2bed8a60 100644
--- a/litellm/llms/vertex_ai/vertex_embeddings/transformation.py
+++ b/litellm/llms/vertex_ai/vertex_embeddings/transformation.py
@@ -1,15 +1,14 @@
 import types
 from typing import Final, Literal
 
-from pydantic import BaseModel
-
 from litellm.llms.vertex_ai.common_utils import pop_vertex_request_labels
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import EmbeddingResponse, Usage
 
 from .types import *
 
 
-class VertexAITextEmbeddingConfig(BaseModel):
+class VertexAITextEmbeddingConfig(LiteLLMBaseModel):
     """
     Reference: https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/text-embeddings-api#TextEmbeddingInput
 
diff --git a/litellm/llms/xai/audio_transcription/transformation.py b/litellm/llms/xai/audio_transcription/transformation.py
index 5d810712634..e34034b5ae9 100644
--- a/litellm/llms/xai/audio_transcription/transformation.py
+++ b/litellm/llms/xai/audio_transcription/transformation.py
@@ -6,11 +6,12 @@ from collections.abc import Mapping, Sequence
 from typing import Final
 
 from httpx import Headers, Response
-from pydantic import BaseModel, ConfigDict, TypeAdapter, ValidationError
+from pydantic import ConfigDict, TypeAdapter, ValidationError
 
 import litellm
 from litellm.litellm_core_utils.audio_utils.utils import process_audio_file
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import (
     AllMessageValues,
     OpenAIAudioTranscriptionOptionalParams,
@@ -28,7 +29,7 @@ class XAIAudioTranscriptionError(BaseLLMException):
     pass
 
 
-class _XAISttWord(BaseModel):
+class _XAISttWord(LiteLLMBaseModel):
     model_config = ConfigDict(extra="allow")
     text: str = ""
     start: float = 0.0
@@ -36,7 +37,7 @@ class _XAISttWord(BaseModel):
     speaker: int | None = None
 
 
-class _XAISttResponse(BaseModel):
+class _XAISttResponse(LiteLLMBaseModel):
     model_config = ConfigDict(extra="allow")
     text: str = ""
     language: str = "unknown"
diff --git a/litellm/llms/xai/batches/transformation.py b/litellm/llms/xai/batches/transformation.py
index 2d986ab99df..44a13ff9c84 100644
--- a/litellm/llms/xai/batches/transformation.py
+++ b/litellm/llms/xai/batches/transformation.py
@@ -15,7 +15,7 @@ import httpx
 from openai.types.batch import BatchRequestCounts
 from openai.types.batch import Errors as BatchErrors
 from openai.types.batch_error import BatchError
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 from typing_extensions import NotRequired, ReadOnly, TypedDict
 
 from litellm.constants import XAI_API_BASE
@@ -23,6 +23,7 @@ from litellm.litellm_core_utils.url_utils import encode_url_path_segment
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
 from litellm.llms.xai.common_utils import XAIModelInfo
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import CreateBatchRequest
 from litellm.types.utils import LiteLLMBatch
 
@@ -89,7 +90,7 @@ class XAICreateBatchRequest(TypedDict):
     input_file_id: NotRequired[ReadOnly[str]]
 
 
-class XAIBatchState(BaseModel):
+class XAIBatchState(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     num_requests: int = 0
@@ -99,7 +100,7 @@ class XAIBatchState(BaseModel):
     num_cancelled: int = 0
 
 
-class XAIBatch(BaseModel):
+class XAIBatch(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     batch_id: str
@@ -112,35 +113,35 @@ class XAIBatch(BaseModel):
     input_file_id: str | None = None
 
 
-class XAIBatchList(BaseModel):
+class XAIBatchList(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     batches: tuple[XAIBatch, ...] = ()
     pagination_token: str | None = None
 
 
-class XAIBatchResultError(BaseModel):
+class XAIBatchResultError(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     code: int | str | None = None
     message: str = ""
 
 
-class XAIBatchResultData(BaseModel):
+class XAIBatchResultData(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     response: Mapping[str, Mapping[str, object]] | None = None
     error: XAIBatchResultError | None = None
 
 
-class XAIBatchResult(BaseModel):
+class XAIBatchResult(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     batch_request_id: str
     batch_result: XAIBatchResultData = XAIBatchResultData()
 
 
-class XAIBatchResultsPage(BaseModel):
+class XAIBatchResultsPage(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     results: tuple[XAIBatchResult, ...] = ()
@@ -202,7 +203,7 @@ def to_litellm_batch(batch: XAIBatch, endpoint: str = DEFAULT_BATCH_ENDPOINT) ->
     )
 
 
-class OpenAIBatchListResponse(BaseModel):
+class OpenAIBatchListResponse(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     object: Literal["list"] = "list"
diff --git a/litellm/llms/xai/files/transformation.py b/litellm/llms/xai/files/transformation.py
index 94fa681d319..33baca79005 100644
--- a/litellm/llms/xai/files/transformation.py
+++ b/litellm/llms/xai/files/transformation.py
@@ -10,13 +10,14 @@ from typing import Final
 
 import httpx
 from openai.types.file_deleted import FileDeleted
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 from typing_extensions import ReadOnly, TypedDict
 
 from litellm.litellm_core_utils.prompt_templates.common_utils import extract_file_data
 from litellm.litellm_core_utils.url_utils import encode_url_path_segment
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
 from litellm.llms.base_llm.files.transformation import BaseFilesConfig, LiteLLMLoggingObj
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import (
     CreateFileRequest,
     FileContentRequest,
@@ -43,7 +44,7 @@ class XAIMultipartUpload(TypedDict):
     purpose: ReadOnly[tuple[None, str]]
 
 
-class XAIFile(BaseModel):
+class XAIFile(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     id: str
@@ -54,14 +55,14 @@ class XAIFile(BaseModel):
     expires_at: int | None = None
 
 
-class XAIFileList(BaseModel):
+class XAIFileList(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     data: tuple[XAIFile, ...] = ()
     pagination_token: str | None = None
 
 
-class XAIFileDeleted(BaseModel):
+class XAIFileDeleted(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     id: str
diff --git a/litellm/main.py b/litellm/main.py
index d81bb75bc46..5b0af2e479c 100644
--- a/litellm/main.py
+++ b/litellm/main.py
@@ -44,6 +44,7 @@ import litellm
 
 # client must be imported from litellm as it's a decorator used at function definition time
 from litellm import client
+from litellm.types.llms.base import LiteLLMBaseModel
 
 # Other utils are imported directly to avoid circular imports
 from litellm.utils import (
@@ -862,11 +863,11 @@ async def _sleep_for_timeout_async(timeout: float | str | httpx.Timeout):
         await asyncio.sleep(timeout.connect)
 
 
-class _AdmissionReservation(BaseModel):
+class _AdmissionReservation(LiteLLMBaseModel):
     input_tokens: int | None = None
 
 
-class _AdmissionMetadata(BaseModel):
+class _AdmissionMetadata(LiteLLMBaseModel):
     user_api_key_budget_reservation: _AdmissionReservation | None = None
 
 
diff --git a/litellm/models/base.py b/litellm/models/base.py
index 8125bfd0205..3105fc9b91d 100644
--- a/litellm/models/base.py
+++ b/litellm/models/base.py
@@ -5,10 +5,12 @@ Base model class for domain models.
 from datetime import datetime
 from typing import Any
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class DomainModel(BaseModel):
+class DomainModel(LiteLLMBaseModel):
     """Base class for all domain models."""
 
     model_config = ConfigDict(
diff --git a/litellm/models/credentials.py b/litellm/models/credentials.py
index e96e406d98e..b91ced275ff 100644
--- a/litellm/models/credentials.py
+++ b/litellm/models/credentials.py
@@ -7,10 +7,12 @@ layer; ``litellm.types.utils`` re-exports them for backwards compatibility.
 
 from collections.abc import Mapping
 
-from pydantic import BaseModel, Field, model_validator
+from pydantic import Field, model_validator
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class CredentialBase(BaseModel):
+class CredentialBase(LiteLLMBaseModel):
     credential_name: str
     credential_info: dict
 
@@ -35,7 +37,7 @@ class CreateCredentialItem(CredentialBase):
         return values
 
 
-class UpdateCredentialItem(BaseModel):
+class UpdateCredentialItem(LiteLLMBaseModel):
     credential_name: str
     credential_info: Mapping[str, object]
     credential_values: Mapping[str, object] | None = None
diff --git a/litellm/models/user.py b/litellm/models/user.py
index 92aca87d303..7dcaed9f77f 100644
--- a/litellm/models/user.py
+++ b/litellm/models/user.py
@@ -7,13 +7,13 @@ Canonical definition for ``litellm_usertable``. Re-exported from
 
 from datetime import datetime
 
-from pydantic import BaseModel, ConfigDict, Field, model_validator
+from pydantic import ConfigDict, Field, model_validator
 
 from litellm.models.object_permission import LiteLLM_ObjectPermissionTable
 from litellm.models.organization_membership import (
     LiteLLM_OrganizationMembershipTable,
 )
-from litellm.types.llms.base import LiteLLMPydanticObjectBase
+from litellm.types.llms.base import LiteLLMBaseModel, LiteLLMPydanticObjectBase
 
 
 class LiteLLM_UserTable(LiteLLMPydanticObjectBase):
@@ -71,7 +71,7 @@ class LiteLLM_UserTable(LiteLLMPydanticObjectBase):
         return model_name in self.models
 
 
-class SCIMPlaceholder(BaseModel):
+class SCIMPlaceholder(LiteLLMBaseModel):
     """A user row keyed by a value that names another account by SSO identity or email."""
 
     placeholder_user_id: str
diff --git a/litellm/proxy/_experimental/mcp_server/discoverable_endpoints.py b/litellm/proxy/_experimental/mcp_server/discoverable_endpoints.py
index 342c650d2a6..26b2d296713 100644
--- a/litellm/proxy/_experimental/mcp_server/discoverable_endpoints.py
+++ b/litellm/proxy/_experimental/mcp_server/discoverable_endpoints.py
@@ -12,7 +12,7 @@ from urllib.parse import parse_qsl, urlencode, urlparse, urlunparse
 import httpx
 from fastapi import APIRouter, Depends, Form, HTTPException, Request
 from fastapi.responses import HTMLResponse, JSONResponse, RedirectResponse, Response
-from pydantic import BaseModel, ConfigDict, Field, SecretStr, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, SecretStr, TypeAdapter, ValidationError
 
 from litellm._logging import verbose_logger
 from litellm.caching.in_memory_cache import InMemoryCache
@@ -91,6 +91,7 @@ from litellm.proxy.common_utils.encrypt_decrypt_utils import (
     encrypt_value_helper,
 )
 from litellm.proxy.common_utils.http_parsing_utils import _read_request_body
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.mcp import MCPAuth, MCPCredentials
 from litellm.types.mcp_server.mcp_server_manager import MCPServer, MCPTokenEndpointAuthMethod
 
@@ -278,7 +279,7 @@ def decode_state_hash(encrypted_state: str) -> dict:
 _BRIDGE_AUTH_CODE_PREFIX: Final = "llm_bcode_"
 
 
-class _BridgeAuthorizationCode(BaseModel):
+class _BridgeAuthorizationCode(LiteLLMBaseModel):
     """Authenticated caller and upstream code sealed for bridge or identity-bound per-user OAuth."""
 
     model_config = ConfigDict(frozen=True)
@@ -340,7 +341,7 @@ def open_bridge_authorization_code(code: str) -> _BridgeAuthorizationCode | None
 _PASSTHROUGH_AUTH_CODE_PREFIX: Final = "llm_ptcode_"
 
 
-class PassthroughAuthorizationCode(BaseModel):
+class PassthroughAuthorizationCode(LiteLLMBaseModel):
     """The ephemeral DCR client and upstream code the gateway seals into the authorization code it
     forwards for a client-forwarded-token server (``true_passthrough`` / ``oauth_delegate``) whose
     authorize fell through to gateway-side registration. These modes forbid the gateway from storing
@@ -1380,7 +1381,7 @@ async def exchange_token_with_server(
     return JSONResponse(result, headers=TOKEN_NO_CACHE_HEADERS)
 
 
-class _DcrClientRegistration(BaseModel):
+class _DcrClientRegistration(LiteLLMBaseModel):
     """RFC 7591 dynamic client registration response, narrowed to the fields the gateway
     must persist to authenticate later token-endpoint calls. Extra members are ignored."""
 
@@ -1389,7 +1390,7 @@ class _DcrClientRegistration(BaseModel):
     token_endpoint_auth_method: str | None = None
 
 
-class _PersistedDcrCredentials(BaseModel):
+class _PersistedDcrCredentials(LiteLLMBaseModel):
     dcr_issuer: str | None = None
     dcr_server_url: str | None = None
     client_id: str | None = None
@@ -1784,7 +1785,7 @@ async def _post_dcr_registration(
     return response
 
 
-class EphemeralDcrClient(BaseModel):
+class EphemeralDcrClient(LiteLLMBaseModel):
     """A DCR client minted for a single authorize round trip and never stored by the gateway."""
 
     model_config = ConfigDict(frozen=True)
diff --git a/litellm/proxy/_experimental/mcp_server/faults/list_outcomes.py b/litellm/proxy/_experimental/mcp_server/faults/list_outcomes.py
index b96a7a74e4a..06bf56f6f96 100644
--- a/litellm/proxy/_experimental/mcp_server/faults/list_outcomes.py
+++ b/litellm/proxy/_experimental/mcp_server/faults/list_outcomes.py
@@ -16,7 +16,7 @@ from typing import Final, Literal, NamedTuple, NoReturn, TypeAlias
 import httpx
 import httpx2
 from mcp.types import Tool as MCPTool
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 from typing_extensions import assert_never
 
 from litellm.proxy._experimental.mcp_server.exceptions import (
@@ -24,6 +24,7 @@ from litellm.proxy._experimental.mcp_server.exceptions import (
     MCPUpstreamAuthError,
 )
 from litellm.proxy._experimental.mcp_server.faults.traversal import iter_exception_tree
+from litellm.types.llms.base import LiteLLMBaseModel
 
 ListFaultCategory: TypeAlias = Literal[
     "auth_required",
@@ -35,13 +36,13 @@ ListFaultCategory: TypeAlias = Literal[
 ]
 
 
-class ServerListOk(BaseModel):
+class ServerListOk(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     tag: Literal["ok"] = "ok"
     tool_count: int
 
 
-class ServerListFault(BaseModel):
+class ServerListFault(LiteLLMBaseModel):
     """Why a server contributed nothing to a listing: the caller must authenticate upstream
     (``auth_required``/``forbidden``), the upstream did not answer (``timeout``/``unreachable``),
     the upstream answered outside its contract (``upstream_error``), or the gateway itself failed
diff --git a/litellm/proxy/_experimental/mcp_server/faults/types.py b/litellm/proxy/_experimental/mcp_server/faults/types.py
index d081d9d735e..e6a52edcf2c 100644
--- a/litellm/proxy/_experimental/mcp_server/faults/types.py
+++ b/litellm/proxy/_experimental/mcp_server/faults/types.py
@@ -9,7 +9,9 @@ from __future__ import annotations
 
 from typing import Final, Literal, TypeAlias
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 MAX_WIRE_FIELD_CHARS: Final = 500
 """Bound on every upstream-derived string that crosses to a caller or into a log line."""
@@ -35,7 +37,7 @@ UPSTREAM_FAULT_CODES: Final[frozenset[str]] = frozenset({"server_error", "tempor
 they classify as upstream-reported faults and render on the 5xx their meaning implies."""
 
 
-class CallerRejected(BaseModel):
+class CallerRejected(LiteLLMBaseModel):
     """The upstream spoke the OAuth error contract and the failure is actionable by our caller
     (e.g. ``invalid_grant``: re-run authorization). The code and its bounded prose relay on the
     4xx status the code itself implies."""
@@ -47,7 +49,7 @@ class CallerRejected(BaseModel):
     error_uri: str | None = None
 
 
-class GatewayRejected(BaseModel):
+class GatewayRejected(LiteLLMBaseModel):
     """The upstream rejected the request for a cause only the gateway operator can address: the
     server's stored client credentials or a gateway capability gap. Not actionable by the caller:
     rendered as 502 with gateway-authored prose naming the code; the upstream's prose goes to
@@ -58,7 +60,7 @@ class GatewayRejected(BaseModel):
     code: str
 
 
-class UpstreamReportedFault(BaseModel):
+class UpstreamReportedFault(LiteLLMBaseModel):
     """The upstream blamed itself in the OAuth vocabulary. Rendered on the 5xx the code implies
     (``server_error`` 502, ``temporarily_unavailable`` 503) so blame and status agree."""
 
@@ -67,7 +69,7 @@ class UpstreamReportedFault(BaseModel):
     code: Literal["server_error", "temporarily_unavailable"]
 
 
-class UpstreamProtocolFault(BaseModel):
+class UpstreamProtocolFault(LiteLLMBaseModel):
     """The upstream broke the error contract: no JSON ``error`` field, an undecodable body, or a
     success response without a usable token. Rendered as 502 with a gateway-authored note; the
     upstream body never crosses to the caller."""
@@ -77,7 +79,7 @@ class UpstreamProtocolFault(BaseModel):
     note: str
 
 
-class UpstreamRegistrationRefused(BaseModel):
+class UpstreamRegistrationRefused(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     tag: Literal["upstream_registration_refused"] = "upstream_registration_refused"
     status_code: Literal[401, 403]
diff --git a/litellm/proxy/_experimental/mcp_server/gateway_dcr_flow.py b/litellm/proxy/_experimental/mcp_server/gateway_dcr_flow.py
index 1136410fd18..c4777179256 100644
--- a/litellm/proxy/_experimental/mcp_server/gateway_dcr_flow.py
+++ b/litellm/proxy/_experimental/mcp_server/gateway_dcr_flow.py
@@ -92,6 +92,7 @@ from litellm.proxy.common_utils.encrypt_decrypt_utils import (
 from litellm.proxy.common_utils.html_forms.native_client_consent import (
     render_native_client_consent_page,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.mcp_server.mcp_server_manager import MCPServer
 
 _DCR_CLAIMS_TARGET: Final = "mcp_dcr_claims"
@@ -172,7 +173,7 @@ credential that LLM routes accept, instead of the MCP-only session pair."""
 ProxyCredentialMintFailure = Literal[ReloadUserFailure, "not_a_member", "team_required"]
 
 
-class MintedProxyCredential(BaseModel):
+class MintedProxyCredential(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     key: str = Field(min_length=1)
     expires_in: int = Field(gt=0)
@@ -216,13 +217,13 @@ SUBJECT_TOKEN_TYPES: Final = frozenset(
 )
 
 
-class SubjectIdentity(BaseModel):
+class SubjectIdentity(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     user_id: str = Field(min_length=1)
     team_id: str | None = None
 
 
-class SubjectTokenRefusal(BaseModel):
+class SubjectTokenRefusal(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     error: Literal["unsupported_grant_type", "invalid_request", "temporarily_unavailable"]
     description: str = Field(min_length=1)
@@ -236,7 +237,7 @@ class ExchangeSubjectToken(Protocol):
     def __call__(self, subject_token: str, request: Request, /) -> Awaitable[SubjectIdentity | SubjectTokenRefusal]: ...
 
 
-class ConsentTeam(BaseModel):
+class ConsentTeam(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     team_id: str = Field(min_length=1)
     team_alias: str | None = None
@@ -276,7 +277,7 @@ async def _unreachable_server(user_id: str, server_id: str) -> bool:
     return False
 
 
-class GatewayDcrClient(BaseModel):
+class GatewayDcrClient(LiteLLMBaseModel):
     """The registration record sealed into a gateway DCR ``client_id``.
 
     ``extra="forbid"`` so a sealed value of another type (an auth code, a connect flow)
@@ -289,7 +290,7 @@ class GatewayDcrClient(BaseModel):
     iat: int
 
 
-class _ConnectFlow(BaseModel):
+class _ConnectFlow(LiteLLMBaseModel):
     """One in-flight authorize: the SSO user it belongs to and the client parameters
     needed to mint the code at the finish step. Sealed into the per-flow cookie. ``jti``
     makes the flow single-use at complete; ``extra="forbid"`` rejects cross-type
@@ -307,7 +308,7 @@ class _ConnectFlow(BaseModel):
     audience: SessionAudience | None = None
 
 
-class _GatewayAuthCode(BaseModel):
+class _GatewayAuthCode(LiteLLMBaseModel):
     """The gateway-sealed authorization code: the user consent it represents and the
     bindings the token endpoint must verify (client, redirect URI, PKCE challenge),
     plus a ``jti`` for the single-use guard. ``extra="forbid"`` rejects cross-type
diff --git a/litellm/proxy/_experimental/mcp_server/mcp_server_manager.py b/litellm/proxy/_experimental/mcp_server/mcp_server_manager.py
index abfb9dc56fe..4e7054cab98 100644
--- a/litellm/proxy/_experimental/mcp_server/mcp_server_manager.py
+++ b/litellm/proxy/_experimental/mcp_server/mcp_server_manager.py
@@ -48,7 +48,7 @@ from mcp.types import (
     ResourceTemplate,
 )
 from mcp.types import Tool as MCPTool
-from pydantic import AnyUrl, BaseModel, TypeAdapter
+from pydantic import AnyUrl, BaseModel, Field, TypeAdapter
 from typing_extensions import ReadOnly
 
 import litellm
@@ -197,6 +197,7 @@ from litellm.proxy.middleware.per_request_root_path_middleware import (
 from litellm.proxy.utils import PrismaClient, ProxyLogging
 from litellm.repositories.table_repositories import MCPServerRepository
 from litellm.types.integrations.slack_alerting import AlertType
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.custom_http import httpxSpecialProvider
 from litellm.types.mcp import (
     DEFAULT_SUBJECT_TOKEN_TYPE,
@@ -223,13 +224,11 @@ try:
         validate_tool_name,  # pyright: ignore[reportAssignmentType]
     )
 except ImportError:
-    from pydantic import BaseModel
-
     SEP_986_URL = "https://github.com/modelcontextprotocol/protocol/blob/main/proposals/0001-tool-name-validation.md"
 
-    class _ToolNameValidationResult(BaseModel):
+    class _ToolNameValidationResult(LiteLLMBaseModel):
         is_valid: bool = True
-        warnings: list = []
+        warnings: list = Field(default=[])
 
     def validate_tool_name(name: str) -> _ToolNameValidationResult:
         return _ToolNameValidationResult()
diff --git a/litellm/proxy/_experimental/mcp_server/outbound_credentials/bridge_credentials.py b/litellm/proxy/_experimental/mcp_server/outbound_credentials/bridge_credentials.py
index ab5fa65480e..1441d9662be 100644
--- a/litellm/proxy/_experimental/mcp_server/outbound_credentials/bridge_credentials.py
+++ b/litellm/proxy/_experimental/mcp_server/outbound_credentials/bridge_credentials.py
@@ -14,7 +14,7 @@ from datetime import datetime
 from functools import lru_cache
 from typing import Final, Literal, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, SecretStr
+from pydantic import ConfigDict, SecretStr
 
 from litellm.proxy._experimental.mcp_server.outbound_credentials.envelope import (
     EnvelopeIdentity,
@@ -32,6 +32,7 @@ from litellm.proxy._experimental.mcp_server.outbound_credentials.envelope import
     open_envelope,
     open_refresh_envelope,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 _SIGNING_KEY_DOMAIN: Final = b"litellm-mcp-bridge:envelope-signing:"
 _ENCRYPTION_KEY_DOMAIN: Final = b"litellm-mcp-bridge:envelope-encryption:"
@@ -110,7 +111,7 @@ def build_bridge_refresh_token_response(
     return mint_refresh_envelope(identity, refresh, keys, now)
 
 
-class BridgeRefreshOpened(BaseModel):
+class BridgeRefreshOpened(LiteLLMBaseModel):
     """A valid refresh envelope presented to the token endpoint: the identity to re-validate and renew
     under, and the upstream refresh grant to exchange."""
 
@@ -120,7 +121,7 @@ class BridgeRefreshOpened(BaseModel):
     refresh: RefreshCredential
 
 
-class BridgeRefreshInvalid(BaseModel):
+class BridgeRefreshInvalid(LiteLLMBaseModel):
     """The presented refresh grant is not a valid refresh envelope for this server (not refresh-shaped,
     will not open, or minted for a different server); the token endpoint fails the refresh closed."""
 
@@ -158,14 +159,14 @@ def open_bridge_refresh_envelope(
     return BridgeRefreshOpened(identity=opened.identity, refresh=opened.refresh)
 
 
-class NotBridgeEnvelope(BaseModel):
+class NotBridgeEnvelope(LiteLLMBaseModel):
     """The bearer is not an envelope; admission continues on its normal path."""
 
     model_config = ConfigDict(frozen=True)
     tag: Literal["not_bridge_envelope"] = "not_bridge_envelope"
 
 
-class BridgeEnvelopeAdmitted(BaseModel):
+class BridgeEnvelopeAdmitted(LiteLLMBaseModel):
     """A valid envelope: the identity to admit under and the full upstream ``Authorization``
     value (``token_type access_token``) to forward to the upstream MCP server."""
 
@@ -175,7 +176,7 @@ class BridgeEnvelopeAdmitted(BaseModel):
     upstream_authorization: SecretStr
 
 
-class BridgeEnvelopeInvalid(BaseModel):
+class BridgeEnvelopeInvalid(LiteLLMBaseModel):
     """The bearer is envelope-shaped but did not open (expired, tampered, wrong key);
     admission must fail closed rather than fall through to normal validation."""
 
diff --git a/litellm/proxy/_experimental/mcp_server/outbound_credentials/client_credentials.py b/litellm/proxy/_experimental/mcp_server/outbound_credentials/client_credentials.py
index 3a8e2b3840a..e628b3880b3 100644
--- a/litellm/proxy/_experimental/mcp_server/outbound_credentials/client_credentials.py
+++ b/litellm/proxy/_experimental/mcp_server/outbound_credentials/client_credentials.py
@@ -35,7 +35,7 @@ from typing import Annotated, Final, Literal
 
 import httpx
 import httpx2
-from pydantic import BaseModel, ConfigDict, Field, SecretStr, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, SecretStr, TypeAdapter, ValidationError
 from typing_extensions import assert_never
 
 from litellm._logging import verbose_logger
@@ -54,9 +54,10 @@ from litellm.proxy._experimental.mcp_server.outbound_credentials.types import (
     CredError,
     HeaderCarrier,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class TokenEndpointSuccess(BaseModel):
+class TokenEndpointSuccess(LiteLLMBaseModel):
     """The endpoint returned a JSON object; field validation is the caller's job."""
 
     model_config = ConfigDict(frozen=True)
@@ -64,7 +65,7 @@ class TokenEndpointSuccess(BaseModel):
     body: dict[str, object]
 
 
-class TokenEndpointDenied(BaseModel):
+class TokenEndpointDenied(LiteLLMBaseModel):
     """The endpoint answered but did not grant a token (an HTTP error or a non-JSON body)."""
 
     model_config = ConfigDict(frozen=True)
@@ -73,7 +74,7 @@ class TokenEndpointDenied(BaseModel):
     detail: str
 
 
-class TokenEndpointUnreachable(BaseModel):
+class TokenEndpointUnreachable(LiteLLMBaseModel):
     """The endpoint could not be reached (DNS, TLS, connect/read failure)."""
 
     model_config = ConfigDict(frozen=True)
diff --git a/litellm/proxy/_experimental/mcp_server/outbound_credentials/envelope.py b/litellm/proxy/_experimental/mcp_server/outbound_credentials/envelope.py
index df883d5a208..7dc5c69994b 100644
--- a/litellm/proxy/_experimental/mcp_server/outbound_credentials/envelope.py
+++ b/litellm/proxy/_experimental/mcp_server/outbound_credentials/envelope.py
@@ -38,9 +38,10 @@ from datetime import datetime, timedelta
 from typing import Final, Literal, TypeAlias
 
 import jwt
-from pydantic import BaseModel, ConfigDict, Field, SecretStr, ValidationError
+from pydantic import ConfigDict, Field, SecretStr, ValidationError
 
 from litellm.proxy.common_utils.encrypt_decrypt_utils import decrypt_value, encrypt_value
+from litellm.types.llms.base import LiteLLMBaseModel
 
 ENVELOPE_PREFIX: Final = "llm_env_"
 """Marker prefix on every serialized ACCESS envelope so the edge can cheaply tell an envelope
@@ -98,7 +99,7 @@ runs both through the same live-policy gate, so team/org/budget/revocation enfor
 identical either way."""
 
 
-class EnvelopeIdentity(BaseModel):
+class EnvelopeIdentity(LiteLLMBaseModel):
     """The litellm principal the envelope binds the inner grant to.
 
     ``subject`` is the principal identifier and ``subject_type`` says how to resolve it: a
@@ -125,7 +126,7 @@ def user_identity(server_id: str, user_id: str) -> EnvelopeIdentity:
     return EnvelopeIdentity(server_id=server_id, subject_type="user_id", subject=user_id)
 
 
-class UpstreamTokenGrant(BaseModel):
+class UpstreamTokenGrant(LiteLLMBaseModel):
     """The upstream OAuth token response fields sealed inside the envelope.
 
     ``expires_in`` must be positive when present; a non-positive value is a programmer
@@ -141,7 +142,7 @@ class UpstreamTokenGrant(BaseModel):
     expires_in: int | None = Field(default=None, gt=0)
 
 
-class RefreshCredential(BaseModel):
+class RefreshCredential(LiteLLMBaseModel):
     """The upstream refresh grant sealed inside a refresh envelope.
 
     Only the refresh token (plus the scope to re-request and the refresh token's own lifetime, when the
@@ -156,7 +157,7 @@ class RefreshCredential(BaseModel):
     expires_in: int | None = Field(default=None, gt=0)
 
 
-class EnvelopeKeys(BaseModel):
+class EnvelopeKeys(LiteLLMBaseModel):
     """Injected key material: the HS256 signing key and the symmetric encryption key.
 
     ``signing_key`` must be at least 32 bytes: HS256's HMAC-SHA256 has a 256-bit
@@ -169,7 +170,7 @@ class EnvelopeKeys(BaseModel):
     encryption_key: SecretStr = Field(min_length=1)
 
 
-class SealedEnvelope(BaseModel):
+class SealedEnvelope(LiteLLMBaseModel):
     """A minted envelope: the client-held bearer value and when it expires."""
 
     model_config = ConfigDict(frozen=True)
@@ -177,7 +178,7 @@ class SealedEnvelope(BaseModel):
     expires_at: datetime
 
 
-class OpenedEnvelope(BaseModel):
+class OpenedEnvelope(LiteLLMBaseModel):
     """A validated access envelope: the identity it was minted for and the recovered grant."""
 
     model_config = ConfigDict(frozen=True)
@@ -185,7 +186,7 @@ class OpenedEnvelope(BaseModel):
     grant: UpstreamTokenGrant
 
 
-class OpenedRefreshEnvelope(BaseModel):
+class OpenedRefreshEnvelope(LiteLLMBaseModel):
     """A validated refresh envelope: the identity it was minted for and the recovered refresh grant."""
 
     model_config = ConfigDict(frozen=True)
@@ -193,7 +194,7 @@ class OpenedRefreshEnvelope(BaseModel):
     refresh: RefreshCredential
 
 
-class EnvelopeTooLarge(BaseModel):
+class EnvelopeTooLarge(LiteLLMBaseModel):
     """The serialized envelope exceeded ``MAX_ENVELOPE_BYTES``; carries sizes only."""
 
     model_config = ConfigDict(frozen=True)
@@ -202,7 +203,7 @@ class EnvelopeTooLarge(BaseModel):
     max_bytes: int
 
 
-class EnvelopeLifetimeUnrepresentable(BaseModel):
+class EnvelopeLifetimeUnrepresentable(LiteLLMBaseModel):
     """A positive provider lifetime cannot be represented as a Python datetime."""
 
     model_config = ConfigDict(frozen=True)
@@ -213,28 +214,28 @@ class EnvelopeLifetimeUnrepresentable(BaseModel):
 EnvelopeMintError: TypeAlias = EnvelopeTooLarge | EnvelopeLifetimeUnrepresentable
 
 
-class NotAnEnvelope(BaseModel):
+class NotAnEnvelope(LiteLLMBaseModel):
     """The candidate does not carry the envelope prefix."""
 
     model_config = ConfigDict(frozen=True)
     tag: Literal["not_an_envelope"] = "not_an_envelope"
 
 
-class BadSignature(BaseModel):
+class BadSignature(LiteLLMBaseModel):
     """The JWT signature does not verify under the provided signing key."""
 
     model_config = ConfigDict(frozen=True)
     tag: Literal["bad_signature"] = "bad_signature"
 
 
-class Expired(BaseModel):
+class Expired(LiteLLMBaseModel):
     """The envelope's ``exp`` is not in the future relative to the provided ``now``."""
 
     model_config = ConfigDict(frozen=True)
     tag: Literal["expired"] = "expired"
 
 
-class MalformedPayload(BaseModel):
+class MalformedPayload(LiteLLMBaseModel):
     """The token is not a well-formed envelope: undecodable JWT, wrong issuer, missing
     or mistyped claims, or a decrypted grant that fails validation."""
 
@@ -242,7 +243,7 @@ class MalformedPayload(BaseModel):
     tag: Literal["malformed_payload"] = "malformed_payload"
 
 
-class DecryptFailed(BaseModel):
+class DecryptFailed(LiteLLMBaseModel):
     """The signed ``grant`` blob could not be decrypted under the provided key."""
 
     model_config = ConfigDict(frozen=True)
@@ -252,7 +253,7 @@ class DecryptFailed(BaseModel):
 EnvelopeOpenError: TypeAlias = NotAnEnvelope | BadSignature | Expired | MalformedPayload | DecryptFailed
 
 
-class _EnvelopeClaims(BaseModel):
+class _EnvelopeClaims(LiteLLMBaseModel):
     """Decoded-claims boundary that pins the exact shape :func:`mint_envelope` emits.
 
     ``server_id``/``key_hash`` mirror the ``min_length`` constraints of
@@ -279,7 +280,7 @@ class _EnvelopeClaims(BaseModel):
     grant: str = Field(min_length=1)
 
 
-class _GrantWire(BaseModel):
+class _GrantWire(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     access_token: str
     token_type: str
@@ -288,7 +289,7 @@ class _GrantWire(BaseModel):
     expires_in: int | None = None
 
 
-class _RefreshWire(BaseModel):
+class _RefreshWire(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     refresh_token: str
     scope: str | None = None
diff --git a/litellm/proxy/_experimental/mcp_server/outbound_credentials/session_credentials.py b/litellm/proxy/_experimental/mcp_server/outbound_credentials/session_credentials.py
index df2bbdba345..b8eda703fa8 100644
--- a/litellm/proxy/_experimental/mcp_server/outbound_credentials/session_credentials.py
+++ b/litellm/proxy/_experimental/mcp_server/outbound_credentials/session_credentials.py
@@ -20,7 +20,7 @@ from datetime import datetime
 from functools import lru_cache
 from typing import Final, Literal, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, Field, SecretStr, ValidationError
+from pydantic import ConfigDict, Field, SecretStr, ValidationError
 
 from litellm.proxy._experimental.mcp_server.outbound_credentials.session_token import (
     AsymmetricSessionKeys,
@@ -35,6 +35,7 @@ from litellm.proxy._experimental.mcp_server.outbound_credentials.session_token i
     open_session_refresh_token,
     open_session_token,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 _SESSION_SIGNING_KEY_DOMAIN: Final = b"litellm-mcp-gateway:session-signing:"
 
@@ -71,7 +72,7 @@ def session_keys_from_master_key(master_key: str) -> SessionKeys:
     return SessionKeys(signing_key=SecretStr(signing))
 
 
-class SessionSigningPreviousKey(BaseModel):
+class SessionSigningPreviousKey(LiteLLMBaseModel):
     """One retired key in ``mcp_session_token_signing.previous_public_keys``: its ``kid``
     and the PEM public half (inline or an ``os.environ/`` reference)."""
 
@@ -80,7 +81,7 @@ class SessionSigningPreviousKey(BaseModel):
     public_key: str = Field(min_length=1)
 
 
-class MCPSessionTokenSigningSettings(BaseModel):
+class MCPSessionTokenSigningSettings(LiteLLMBaseModel):
     """The ``general_settings.mcp_session_token_signing`` block: opt-in asymmetric signing
     for the gateway session tokens. Absent, the gateway keeps the backward-compatible
     HS256 key derived from ``master_key``. ``private_key`` and each ``public_key`` accept
@@ -93,7 +94,7 @@ class MCPSessionTokenSigningSettings(BaseModel):
     previous_public_keys: tuple[SessionSigningPreviousKey, ...] = ()
 
 
-class SessionSigningConfigError(BaseModel):
+class SessionSigningConfigError(LiteLLMBaseModel):
     """``mcp_session_token_signing`` is present but unusable (bad shape, unresolvable
     secret reference, or a key that is not a loadable RSA PEM); the caller fails closed
     with a server error instead of silently falling back to HS256."""
@@ -164,14 +165,14 @@ def active_session_signing_keys(master_key: str) -> SessionSigningKeys | Session
     return resolve_session_signing_keys(master_key, general_settings.get("mcp_session_token_signing"))
 
 
-class NotSessionBearer(BaseModel):
+class NotSessionBearer(LiteLLMBaseModel):
     """The bearer is not session-shaped; admission continues on its normal path."""
 
     model_config = ConfigDict(frozen=True)
     tag: Literal["not_session_bearer"] = "not_session_bearer"
 
 
-class SessionBearerAdmitted(BaseModel):
+class SessionBearerAdmitted(LiteLLMBaseModel):
     """A valid session access token: the principal to admit under after a live reload."""
 
     model_config = ConfigDict(frozen=True)
@@ -179,7 +180,7 @@ class SessionBearerAdmitted(BaseModel):
     principal: SessionPrincipal
 
 
-class SessionBearerInvalid(BaseModel):
+class SessionBearerInvalid(LiteLLMBaseModel):
     """The bearer is session-shaped but must not admit (expired, tampered, wrong key, or a
     refresh token presented at the tool-call edge); admission fails closed with the
     ``invalid_token`` challenge rather than falling through to another arm. ``expired``
@@ -238,7 +239,7 @@ def resolve_session_bearer(
     return SessionBearerInvalid(expired=isinstance(opened, SessionExpired))
 
 
-class SessionRefreshOpened(BaseModel):
+class SessionRefreshOpened(LiteLLMBaseModel):
     """A valid session refresh token presented to the token endpoint: the principal to
     re-validate and renew under."""
 
@@ -248,7 +249,7 @@ class SessionRefreshOpened(BaseModel):
     jti: str
 
 
-class SessionRefreshInvalid(BaseModel):
+class SessionRefreshInvalid(LiteLLMBaseModel):
     """The presented refresh grant is not a valid session refresh token for this client
     (not refresh-shaped, will not open, or bound to a different ``client_id``); the token
     endpoint fails the refresh closed."""
diff --git a/litellm/proxy/_experimental/mcp_server/outbound_credentials/session_token.py b/litellm/proxy/_experimental/mcp_server/outbound_credentials/session_token.py
index 0fa750a4c4a..da91faccadd 100644
--- a/litellm/proxy/_experimental/mcp_server/outbound_credentials/session_token.py
+++ b/litellm/proxy/_experimental/mcp_server/outbound_credentials/session_token.py
@@ -43,7 +43,9 @@ import jwt
 from cryptography.exceptions import UnsupportedAlgorithm
 from cryptography.hazmat.primitives import serialization
 from cryptography.hazmat.primitives.asymmetric import rsa
-from pydantic import BaseModel, ConfigDict, Field, SecretStr, ValidationError, field_validator, model_validator
+from pydantic import ConfigDict, Field, SecretStr, ValidationError, field_validator, model_validator
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 SESSION_TOKEN_PREFIX: Final = "llm_session_"
 """Marker prefix on every serialized session ACCESS token so the admission edge can cheaply
@@ -97,7 +99,7 @@ is read only from the signed claims, never from the request, so a token of one a
 never be redeemed as the other."""
 
 
-class SessionPrincipal(BaseModel):
+class SessionPrincipal(LiteLLMBaseModel):
     """The litellm user a session token identifies and the DCR client it was issued to.
 
     ``user_id`` is the SSO-established litellm user subject, never a credential: admission
@@ -121,7 +123,7 @@ class SessionPrincipal(BaseModel):
     team_id: str | None = None
 
 
-class SessionKeys(BaseModel):
+class SessionKeys(LiteLLMBaseModel):
     """Injected key material: the HS256 signing key.
 
     ``signing_key`` must be at least 32 bytes: HS256's HMAC-SHA256 has a 256-bit security
@@ -133,7 +135,7 @@ class SessionKeys(BaseModel):
     signing_key: SecretStr = Field(min_length=32)
 
 
-class SessionRotatedPublicKey(BaseModel):
+class SessionRotatedPublicKey(LiteLLMBaseModel):
     """The public half of a retired signing key, kept verifiable under its ``kid`` during a
     rotation window so tokens minted before the rotation stay valid until they expire."""
 
@@ -155,7 +157,7 @@ class SessionRotatedPublicKey(BaseModel):
         return value
 
 
-class AsymmetricSessionKeys(BaseModel):
+class AsymmetricSessionKeys(LiteLLMBaseModel):
     """Injected RS256 key material: the issuer-held RSA private key and the stable ``kid``
     stamped into every minted token's JOSE header, plus the public halves of previously
     rotated keys that verification still accepts while their tokens age out. Downstream
@@ -212,7 +214,7 @@ def session_public_key_pem(keys: AsymmetricSessionKeys) -> str:
     return _public_key_pem_from_private(keys.private_key_pem.get_secret_value())
 
 
-class MintedSessionToken(BaseModel):
+class MintedSessionToken(LiteLLMBaseModel):
     """A minted session token: the client-held bearer value and when it expires."""
 
     model_config = ConfigDict(frozen=True)
@@ -220,7 +222,7 @@ class MintedSessionToken(BaseModel):
     expires_at: datetime
 
 
-class OpenedSessionToken(BaseModel):
+class OpenedSessionToken(LiteLLMBaseModel):
     """A validated session token of either kind: the principal it was minted for, the
     ``jti`` so the token endpoint can enforce single-use rotation on a refresh token, and
     the signed ``kind``/``iat``/``exp`` so an introspection response can report the
@@ -234,7 +236,7 @@ class OpenedSessionToken(BaseModel):
     exp: int
 
 
-class SessionTokenTooLarge(BaseModel):
+class SessionTokenTooLarge(LiteLLMBaseModel):
     """The serialized token exceeded ``MAX_SESSION_TOKEN_BYTES``; carries sizes only. Only
     reachable through an oversized ``client_id``, which registration should have bounded."""
 
@@ -247,28 +249,28 @@ class SessionTokenTooLarge(BaseModel):
 SessionTokenMintError: TypeAlias = SessionTokenTooLarge
 
 
-class NotASessionToken(BaseModel):
+class NotASessionToken(LiteLLMBaseModel):
     """The candidate does not carry the expected session prefix."""
 
     model_config = ConfigDict(frozen=True)
     tag: Literal["not_a_session_token"] = "not_a_session_token"
 
 
-class SessionBadSignature(BaseModel):
+class SessionBadSignature(LiteLLMBaseModel):
     """The JWT signature does not verify under the provided signing key."""
 
     model_config = ConfigDict(frozen=True)
     tag: Literal["session_bad_signature"] = "session_bad_signature"
 
 
-class SessionExpired(BaseModel):
+class SessionExpired(LiteLLMBaseModel):
     """The token's ``exp`` is not in the future relative to the provided ``now``."""
 
     model_config = ConfigDict(frozen=True)
     tag: Literal["session_expired"] = "session_expired"
 
 
-class SessionMalformed(BaseModel):
+class SessionMalformed(LiteLLMBaseModel):
     """The token is not a well-formed session token: undecodable JWT, wrong issuer, wrong
     ``kind``, or missing/mistyped/extra claims."""
 
@@ -279,7 +281,7 @@ class SessionMalformed(BaseModel):
 SessionTokenOpenError: TypeAlias = NotASessionToken | SessionBadSignature | SessionExpired | SessionMalformed
 
 
-class _SessionClaims(BaseModel):
+class _SessionClaims(LiteLLMBaseModel):
     """Decoded-claims boundary that pins the exact shape the mints emit.
 
     ``user_id``/``client_id`` mirror the ``min_length`` constraints of
@@ -469,7 +471,7 @@ def _open(
     )
 
 
-class _VerificationMaterial(BaseModel):
+class _VerificationMaterial(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     key: SecretStr
     algorithm: Literal["HS256", "RS256"]
diff --git a/litellm/proxy/_experimental/mcp_server/outbound_credentials/sso_assertion_store.py b/litellm/proxy/_experimental/mcp_server/outbound_credentials/sso_assertion_store.py
index 5503d19211b..0bc17d94165 100644
--- a/litellm/proxy/_experimental/mcp_server/outbound_credentials/sso_assertion_store.py
+++ b/litellm/proxy/_experimental/mcp_server/outbound_credentials/sso_assertion_store.py
@@ -23,11 +23,12 @@ from datetime import datetime, timezone
 from typing import TYPE_CHECKING, Final, Protocol
 
 import jwt
-from pydantic import BaseModel, ConfigDict, SecretStr, TypeAdapter, ValidationError
+from pydantic import ConfigDict, SecretStr, TypeAdapter, ValidationError
 
 from litellm._logging import verbose_proxy_logger
 from litellm.caching.in_memory_cache import InMemoryCache
 from litellm.constants import MCP_OAUTH2_TOKEN_CACHE_MAX_SIZE, MCP_SSO_ASSERTION_CACHE_TTL_SECONDS
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from prisma.models import LiteLLM_SSOIdentityAssertion
@@ -67,7 +68,7 @@ def _mcp_server_table(prisma_client: PrismaClient) -> _MCPServerTable:
     return prisma_client.db.litellm_mcpservertable
 
 
-class SSOIdentityAssertion(BaseModel):
+class SSOIdentityAssertion(LiteLLMBaseModel):
     """The IdP material an EMA exchange needs: ``id_token`` is the RFC 8693 subject token,
     ``expires_at`` bounds its usefulness, and the refresh token renews it without re-login."""
 
@@ -119,12 +120,12 @@ class SSOAssertionCache:
 _ASSERTION_CACHE: Final = SSOAssertionCache()
 
 
-class _IdTokenClaims(BaseModel):
+class _IdTokenClaims(LiteLLMBaseModel):
     exp: float | None = None
     iss: str | None = None
 
 
-class _StoredAssertionPayload(BaseModel):
+class _StoredAssertionPayload(LiteLLMBaseModel):
     id_token: str
     refresh_token: str | None = None
     issuer: str | None = None
diff --git a/litellm/proxy/_experimental/mcp_server/outbound_credentials/token_endpoint.py b/litellm/proxy/_experimental/mcp_server/outbound_credentials/token_endpoint.py
index 6f1e5baa109..1ee1b7ecf2f 100644
--- a/litellm/proxy/_experimental/mcp_server/outbound_credentials/token_endpoint.py
+++ b/litellm/proxy/_experimental/mcp_server/outbound_credentials/token_endpoint.py
@@ -24,7 +24,7 @@ from typing import Final
 
 import httpx
 import jwt
-from pydantic import BaseModel, TypeAdapter, ValidationError
+from pydantic import TypeAdapter, ValidationError
 from typing_extensions import assert_never
 
 from litellm._logging import verbose_proxy_logger
@@ -50,6 +50,7 @@ from litellm.proxy._experimental.mcp_server.outbound_credentials.types import (
     CredError,
     PrivateKeyJwtAuth,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.custom_http import httpxSpecialProvider
 
 # The cache stores (fingerprint, token); anything else in the slot is treated as absent.
@@ -65,7 +66,7 @@ class ExchangedToken:
     expires_in: int | None
 
 
-class _TokenEndpointResponse(BaseModel):
+class _TokenEndpointResponse(LiteLLMBaseModel):
     access_token: str
     expires_in: int | None = None
 
diff --git a/litellm/proxy/_experimental/mcp_server/outbound_credentials/types.py b/litellm/proxy/_experimental/mcp_server/outbound_credentials/types.py
index 33c3a854058..55a73687b6f 100644
--- a/litellm/proxy/_experimental/mcp_server/outbound_credentials/types.py
+++ b/litellm/proxy/_experimental/mcp_server/outbound_credentials/types.py
@@ -32,7 +32,7 @@ from typing import Annotated, Final, Literal
 
 import httpx2
 from expression import case, tag, tagged_union
-from pydantic import BaseModel, ConfigDict, Field, SecretStr, field_validator
+from pydantic import ConfigDict, Field, SecretStr, field_validator
 from typing_extensions import assert_never
 
 from litellm.proxy._experimental.mcp_server.outbound_credentials.result import (
@@ -40,6 +40,7 @@ from litellm.proxy._experimental.mcp_server.outbound_credentials.result import (
     Ok,
     Result,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.mcp import (
     DEFAULT_CREDENTIAL_HEADER,
     DEFAULT_SUBJECT_TOKEN_TYPE,
@@ -213,7 +214,7 @@ def validate_header_name(raw: str) -> Result[str, CredError]:
     return Ok(normalized)
 
 
-class HeaderCarrier(BaseModel):
+class HeaderCarrier(LiteLLMBaseModel):
     """Where a resolved credential is written upstream, and how its value is formatted.
 
     ``Authorization: Bearer`` is only OAuth's *default* conveyance (RFC 6750 section 2.1), not its
@@ -319,7 +320,7 @@ class TokenExchangeConfig(HeaderCarrier):
     scopes: tuple[str, ...] = ()
 
 
-class PrivateKeyJwtAuth(BaseModel):
+class PrivateKeyJwtAuth(LiteLLMBaseModel):
     """RFC 7523 private-key-JWT client authentication: the gateway signs a `client_assertion`."""
 
     model_config = ConfigDict(frozen=True)
@@ -329,7 +330,7 @@ class PrivateKeyJwtAuth(BaseModel):
     signing_alg: str = "RS256"
 
 
-class ClientSecretAuth(BaseModel):
+class ClientSecretAuth(LiteLLMBaseModel):
     """`client_secret_post` client authentication: the gateway posts `client_id` + `client_secret`."""
 
     model_config = ConfigDict(frozen=True)
@@ -362,7 +363,7 @@ class IdJagConfig(HeaderCarrier):
     scopes: tuple[str, ...] = ()
 
 
-class SharedKey(BaseModel):
+class SharedKey(LiteLLMBaseModel):
     """A fixed key configured on the server, identical for every caller."""
 
     model_config = ConfigDict(frozen=True)
@@ -370,7 +371,7 @@ class SharedKey(BaseModel):
     value: SecretStr
 
 
-class Byok(BaseModel):
+class Byok(LiteLLMBaseModel):
     """A key the user brings via the entry flow, stored per-user and pulled from the credential
     store at resolve time. Missing means the user must provide it, a 401 + WWW-Authenticate
     challenge."""
@@ -393,21 +394,21 @@ class ApiKeyConfig(HeaderCarrier):
     key_source: ApiKeySource
 
 
-class PassthroughConfig(BaseModel):
+class PassthroughConfig(LiteLLMBaseModel):
     """Client-driven upstream OAuth; the gateway forwards the client's upstream token."""
 
     model_config = ConfigDict(frozen=True)
     kind: Literal[AuthSpecKind.passthrough] = AuthSpecKind.passthrough
 
 
-class NoneConfig(BaseModel):
+class NoneConfig(LiteLLMBaseModel):
     """No upstream credential; the request is sent unauthenticated."""
 
     model_config = ConfigDict(frozen=True)
     kind: Literal[AuthSpecKind.none] = AuthSpecKind.none
 
 
-class StaticKeys(BaseModel):
+class StaticKeys(LiteLLMBaseModel):
     """Long-lived AWS access keys configured on the server."""
 
     model_config = ConfigDict(frozen=True)
@@ -417,7 +418,7 @@ class StaticKeys(BaseModel):
     session_token: SecretStr | None = None
 
 
-class AssumeRole(BaseModel):
+class AssumeRole(LiteLLMBaseModel):
     """An IAM role the gateway assumes via STS for short-lived, auto-refreshed credentials."""
 
     model_config = ConfigDict(frozen=True)
@@ -427,7 +428,7 @@ class AssumeRole(BaseModel):
     external_id: str | None = None
 
 
-class Ambient(BaseModel):
+class Ambient(LiteLLMBaseModel):
     """The environment's default AWS credential chain (instance profile, IRSA, env vars)."""
 
     model_config = ConfigDict(frozen=True)
@@ -437,7 +438,7 @@ class Ambient(BaseModel):
 AwsCredentialSource = Annotated[StaticKeys | AssumeRole | Ambient, Field(discriminator="source")]
 
 
-class AwsSigV4Config(BaseModel):
+class AwsSigV4Config(LiteLLMBaseModel):
     """AWS SigV4 per-request signing for an AWS-hosted upstream (e.g. Bedrock AgentCore). The
     gateway signs with its own AWS identity, never the caller's; `credentials` selects how that
     identity is obtained, defaulting to the ambient credential chain."""
@@ -462,7 +463,7 @@ AuthConfig = Annotated[
 ]
 
 
-class Subject(BaseModel):
+class Subject(LiteLLMBaseModel):
     """The validated inbound principal. NOT the v1 request object and NOT the LiteLLM key."""
 
     model_config = ConfigDict(frozen=True)
@@ -474,7 +475,7 @@ class Subject(BaseModel):
     inbound_token: SecretStr | None = None
 
 
-class ServerSpec(BaseModel):
+class ServerSpec(LiteLLMBaseModel):
     """The declared upstream. A v2-native type; the v1 -> v2 adapter maps onto this."""
 
     model_config = ConfigDict(frozen=True)
diff --git a/litellm/proxy/_types.py b/litellm/proxy/_types.py
index c0f2472a6bd..291e991f6bd 100644
--- a/litellm/proxy/_types.py
+++ b/litellm/proxy/_types.py
@@ -38,6 +38,7 @@ from litellm.types.integrations.otel_span_attributes import (
     SpanAttributes as SpanAttributes,  # noqa: PLC0414  # public re-export
 )
 from litellm.types.integrations.slack_alerting import AlertType
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import (
     AllMessageValues,
     ResponsesAPIResponse,
@@ -2599,7 +2600,7 @@ class CallbackDelete(LiteLLMPydanticObjectBase):
     callback_name: str
 
 
-class FieldDetail(BaseModel):
+class FieldDetail(LiteLLMBaseModel):
     field_name: str
     field_type: str
     field_description: str
@@ -3834,7 +3835,7 @@ class LiteLLM_ProjectTableCachedObj(LiteLLM_ProjectTable):
     last_refreshed_at: float | None = None
 
 
-class LiteLLM_UserTableFiltered(BaseModel):  # done to avoid exposing sensitive data
+class LiteLLM_UserTableFiltered(LiteLLMBaseModel):  # done to avoid exposing sensitive data
     user_id: str
     user_email: str | None = None
 
@@ -4750,14 +4751,14 @@ class TeamMemberUpdateResponse(MemberUpdateResponse):
     temp_budget_expiry: datetime | None = None
 
 
-class TeamModelAddRequest(BaseModel):
+class TeamModelAddRequest(LiteLLMBaseModel):
     """Request to add models to a team"""
 
     team_id: str
     models: list[str]
 
 
-class TeamModelDeleteRequest(BaseModel):
+class TeamModelDeleteRequest(LiteLLMBaseModel):
     """Request to delete models from a team"""
 
     team_id: str
@@ -4812,20 +4813,20 @@ class TeamInfoMember(Member):
     user_alias: str | None = None
 
 
-class TeamEditUnrestricted(BaseModel):
+class TeamEditUnrestricted(LiteLLMBaseModel):
     kind: Literal["unrestricted"] = "unrestricted"
 
 
-class TeamEditAsTeamAdmin(BaseModel):
+class TeamEditAsTeamAdmin(LiteLLMBaseModel):
     kind: Literal["team_admin"] = "team_admin"
     editable_fields: tuple[str, ...]
 
 
-class TeamEditAsTeamAdminDisabled(BaseModel):
+class TeamEditAsTeamAdminDisabled(LiteLLMBaseModel):
     kind: Literal["team_admin_disabled"] = "team_admin_disabled"
 
 
-class TeamEditNone(BaseModel):
+class TeamEditNone(LiteLLMBaseModel):
     kind: Literal["none"] = "none"
 
 
@@ -4864,7 +4865,7 @@ class TeamInfoResponseObject(TypedDict):
     team_memberships: ReadOnly[tuple[TeamInfoMembership, ...]]
 
 
-class TeamMemberResetBudgetResponse(BaseModel):
+class TeamMemberResetBudgetResponse(LiteLLMBaseModel):
     team_id: str
     user_id: str
     budget_id: str | None
@@ -5150,12 +5151,12 @@ class RoleBasedPermissions(OIDCPermissions):
     }
 
 
-class RoleMapping(BaseModel):
+class RoleMapping(LiteLLMBaseModel):
     role: str
     internal_role: RBAC_ROLES
 
 
-class JWTLiteLLMRoleMap(BaseModel):
+class JWTLiteLLMRoleMap(LiteLLMBaseModel):
     jwt_role: str
     litellm_role: LitellmUserRoles
 
@@ -5168,7 +5169,7 @@ class ScopeMapping(OIDCPermissions):
     }
 
 
-class JWTRoutingOverride(BaseModel):
+class JWTRoutingOverride(LiteLLMBaseModel):
     """
     Override default auth routing for JWT-shaped bearer tokens.
 
@@ -5187,9 +5188,7 @@ class JWTRoutingOverride(BaseModel):
     aud: str | list[str] | None = None
     path: Literal["oauth2"] = "oauth2"
 
-    model_config = {
-        "extra": "forbid",
-    }
+    model_config = ConfigDict(extra="forbid")
 
 
 class UnregisteredJWTClientBehavior(str, enum.Enum):
@@ -5211,7 +5210,7 @@ class UnregisteredJWTClientBehavior(str, enum.Enum):
     AUTO_REGISTER = "auto_register"
 
 
-class JWTIssuerConfig(BaseModel):
+class JWTIssuerConfig(LiteLLMBaseModel):
     """
     Issuer-bound JWT validation configuration.
 
@@ -5268,9 +5267,7 @@ class JWTIssuerConfig(BaseModel):
         description="Issuer-specific policy when the virtual key claim has no mapping. Falls back to the global policy.",
     )
 
-    model_config = {
-        "extra": "forbid",
-    }
+    model_config = ConfigDict(extra="forbid")
 
     @model_validator(mode="after")
     def validate_audience_configured(self) -> "JWTIssuerConfig":
@@ -5567,7 +5564,7 @@ class SpecialManagementEndpointEnums(enum.Enum):
     DEFAULT_ORGANIZATION = "default_organization"
 
 
-class TransformRequestBody(BaseModel):
+class TransformRequestBody(LiteLLMBaseModel):
     call_type: CallTypes
     request_body: dict
 
diff --git a/litellm/proxy/a2a/endpoints.py b/litellm/proxy/a2a/endpoints.py
index 57fc5f1dce2..56b512df934 100644
--- a/litellm/proxy/a2a/endpoints.py
+++ b/litellm/proxy/a2a/endpoints.py
@@ -13,7 +13,7 @@ from typing import Any, Final
 
 from fastapi import APIRouter, Depends, HTTPException
 from fastapi.responses import JSONResponse
-from pydantic import BaseModel, Field
+from pydantic import Field
 
 from litellm._logging import verbose_proxy_logger
 from litellm.proxy._types import LitellmUserRoles, UserAPIKeyAuth
@@ -24,11 +24,12 @@ from litellm.proxy.a2a.discovery import (
     fetch_well_known_card,
 )
 from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
+from litellm.types.llms.base import LiteLLMBaseModel
 
 router: Final = APIRouter()
 
 
-class DiscoverAgentRequest(BaseModel):
+class DiscoverAgentRequest(LiteLLMBaseModel):
     url: str = Field(
         ...,
         description=(
@@ -58,7 +59,7 @@ class DiscoverAgentRequest(BaseModel):
     )
 
 
-class DiscoverAgentResponse(BaseModel):
+class DiscoverAgentResponse(LiteLLMBaseModel):
     url: str
     agent_card: dict[str, Any]
 
diff --git a/litellm/proxy/agent_endpoints/agent_search.py b/litellm/proxy/agent_endpoints/agent_search.py
index 65a89bb2c7a..1d1778a960b 100644
--- a/litellm/proxy/agent_endpoints/agent_search.py
+++ b/litellm/proxy/agent_endpoints/agent_search.py
@@ -6,7 +6,7 @@ from collections.abc import Sequence
 from dataclasses import dataclass
 from typing import TYPE_CHECKING, Final, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, ValidationError
+from pydantic import ConfigDict, ValidationError
 
 from litellm.proxy.common_utils.semantic_text_index import (
     Embedder,
@@ -15,6 +15,7 @@ from litellm.proxy.common_utils.semantic_text_index import (
     router_embedder,
 )
 from litellm.types.agents import AgentResponse
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from litellm.proxy._types import UserAPIKeyAuth
@@ -48,7 +49,7 @@ class AgentSearchEmbeddingFailed:
 AgentSearchOutcome: TypeAlias = AgentSearchHits | AgentSearchNotConfigured | AgentSearchEmbeddingFailed
 
 
-class _SearchableSkill(BaseModel):
+class _SearchableSkill(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     name: str = ""
@@ -56,14 +57,14 @@ class _SearchableSkill(BaseModel):
     tags: tuple[str, ...] = ()
 
 
-class _SearchableCard(BaseModel):
+class _SearchableCard(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     description: str = ""
     skills: tuple[_SearchableSkill, ...] = ()
 
 
-class AgentSearchResult(BaseModel):
+class AgentSearchResult(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     agent_id: str
diff --git a/litellm/proxy/analytics_endpoints/cache_activity.py b/litellm/proxy/analytics_endpoints/cache_activity.py
index 5de3b610782..c860f3c8a88 100644
--- a/litellm/proxy/analytics_endpoints/cache_activity.py
+++ b/litellm/proxy/analytics_endpoints/cache_activity.py
@@ -4,9 +4,10 @@ from collections.abc import Sequence
 from datetime import datetime
 from typing import Final, Protocol
 
-from pydantic import BaseModel, TypeAdapter
+from pydantic import TypeAdapter
 
 from litellm.proxy._types import LiteLLMRoutes
+from litellm.types.llms.base import LiteLLMBaseModel
 
 UNKNOWN_CALL_TYPE: Final = "Unknown"
 INFO_ROUTES_JSON: Final = json.dumps(LiteLLMRoutes.info_routes.value)
@@ -25,7 +26,7 @@ class _SupportsRawQueryDb(Protocol):
     def db(self) -> _SupportsQueryRaw: ...
 
 
-class CacheActivityGroup(BaseModel):
+class CacheActivityGroup(LiteLLMBaseModel):
     call_type: str
     api_requests: int
     cache_hits: int
@@ -34,7 +35,7 @@ class CacheActivityGroup(BaseModel):
     generated_completion_tokens: int
 
 
-class CacheActivityTotals(BaseModel):
+class CacheActivityTotals(LiteLLMBaseModel):
     api_requests: int
     cache_hits: int
     failed_requests: int
@@ -42,19 +43,19 @@ class CacheActivityTotals(BaseModel):
     cache_hit_ratio: float
 
 
-class CacheActivityFilterOptions(BaseModel):
+class CacheActivityFilterOptions(LiteLLMBaseModel):
     key_aliases: list[str]
     models: list[str]
 
 
-class CacheActivityErrorBucket(BaseModel):
+class CacheActivityErrorBucket(LiteLLMBaseModel):
     call_type: str
     error_code: str
     error_class: str
     count: int
 
 
-class CacheActivityResponse(BaseModel):
+class CacheActivityResponse(LiteLLMBaseModel):
     groups: list[CacheActivityGroup]
     totals: CacheActivityTotals
     filter_options: CacheActivityFilterOptions
@@ -131,11 +132,11 @@ MODEL_OPTIONS_SQL: Final = """
 """
 
 
-class _KeyAliasRow(BaseModel):
+class _KeyAliasRow(LiteLLMBaseModel):
     key_alias: str
 
 
-class _ModelRow(BaseModel):
+class _ModelRow(LiteLLMBaseModel):
     model: str
 
 
diff --git a/litellm/proxy/anthropic_endpoints/gateway_endpoints.py b/litellm/proxy/anthropic_endpoints/gateway_endpoints.py
index e84bb73b05b..48acfa67a28 100644
--- a/litellm/proxy/anthropic_endpoints/gateway_endpoints.py
+++ b/litellm/proxy/anthropic_endpoints/gateway_endpoints.py
@@ -24,7 +24,7 @@ from typing import Final
 
 from fastapi import APIRouter, Depends, Request, Response
 from fastapi.responses import JSONResponse
-from pydantic import BaseModel, Field, TypeAdapter, ValidationError
+from pydantic import Field, TypeAdapter, ValidationError
 
 from litellm._internal_context import with_service_target
 from litellm._logging import verbose_proxy_logger
@@ -40,6 +40,7 @@ from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
 from litellm.proxy.common_utils.http_parsing_utils import _safe_set_request_parsed_body
 from litellm.proxy.management_endpoints.sso_helper_utils import CLI_SSO_SESSIONS_TARGET
 from litellm.proxy.management_endpoints.ui_sso import CliSsoTeamDetail
+from litellm.types.llms.base import LiteLLMBaseModel
 
 GATEWAY_PREFIX: Final = "/claude_code_gateway"
 _DEVICE_CODE_GRANT: Final = "urn:ietf:params:oauth:grant-type:device_code"
@@ -52,7 +53,7 @@ _NO_SETTINGS: Final = MappingProxyType({})
 _POST_ONLY: Final = ["POST"]
 
 
-class _GatewaySessionData(BaseModel):
+class _GatewaySessionData(LiteLLMBaseModel):
     user_id: str
     user_role: LitellmUserRoles
     models: list[str] = Field(default_factory=list)
@@ -67,19 +68,19 @@ class _GatewayLogin:
     team: CliSsoTeamDetail
 
 
-class _OAuthErrorBody(BaseModel):
+class _OAuthErrorBody(LiteLLMBaseModel):
     error: str
     error_description: str | None = None
 
 
-class _AuthorizationServerMetadata(BaseModel):
+class _AuthorizationServerMetadata(LiteLLMBaseModel):
     issuer: str
     device_authorization_endpoint: str
     token_endpoint: str
     grant_types_supported: tuple[str, ...]
 
 
-class _DeviceAuthorizationBody(BaseModel):
+class _DeviceAuthorizationBody(LiteLLMBaseModel):
     device_code: str
     user_code: str
     verification_uri: str
@@ -88,13 +89,13 @@ class _DeviceAuthorizationBody(BaseModel):
     interval: int
 
 
-class _AccessTokenBody(BaseModel):
+class _AccessTokenBody(LiteLLMBaseModel):
     access_token: str
     expires_in: int
     token_type: str = "Bearer"
 
 
-class _ManagedSettingsBody(BaseModel):
+class _ManagedSettingsBody(LiteLLMBaseModel):
     uuid: str
     checksum: str
     settings: dict[str, object]
diff --git a/litellm/proxy/auth/fallback_budget.py b/litellm/proxy/auth/fallback_budget.py
index e356f8acc7d..9029f9d6996 100644
--- a/litellm/proxy/auth/fallback_budget.py
+++ b/litellm/proxy/auth/fallback_budget.py
@@ -35,7 +35,7 @@ from collections.abc import Callable, Mapping
 from dataclasses import dataclass
 from typing import Final
 
-from pydantic import BaseModel, ValidationError
+from pydantic import ValidationError
 
 from litellm._logging import verbose_proxy_logger
 from litellm.proxy._types import UserAPIKeyAuth
@@ -43,13 +43,14 @@ from litellm.proxy.auth.auth_checks import (
     _is_model_cost_zero,  # pyright: ignore[reportPrivateUsage]  # the zero-cost predicate the auth-time budget checks use; no public equivalent
 )
 from litellm.router import Router
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class _RequestMetadata(BaseModel):
+class _RequestMetadata(LiteLLMBaseModel):
     user_api_key_auth: UserAPIKeyAuth | None = None
 
 
-class _FallbackBudgetSettings(BaseModel):
+class _FallbackBudgetSettings(LiteLLMBaseModel):
     enforce_fallback_budget: bool = True
 
 
diff --git a/litellm/proxy/auth/fallback_model_access.py b/litellm/proxy/auth/fallback_model_access.py
index c601a5e415e..53807f076ea 100644
--- a/litellm/proxy/auth/fallback_model_access.py
+++ b/litellm/proxy/auth/fallback_model_access.py
@@ -12,19 +12,20 @@ from collections.abc import Callable, Mapping
 from dataclasses import dataclass
 from typing import Final
 
-from pydantic import BaseModel, ValidationError
+from pydantic import ValidationError
 
 from litellm._logging import verbose_proxy_logger
 from litellm.proxy._types import ProxyException, UserAPIKeyAuth
 from litellm.proxy.auth.auth_checks import can_key_call_resolved_model
 from litellm.router import Router
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class _RequestMetadata(BaseModel):
+class _RequestMetadata(LiteLLMBaseModel):
     user_api_key_auth: UserAPIKeyAuth | None = None
 
 
-class _FallbackAccessSettings(BaseModel):
+class _FallbackAccessSettings(LiteLLMBaseModel):
     enforce_fallback_model_access: bool = False
 
 
diff --git a/litellm/proxy/auth/network.py b/litellm/proxy/auth/network.py
index 4e8ab7512a7..230dde072ca 100644
--- a/litellm/proxy/auth/network.py
+++ b/litellm/proxy/auth/network.py
@@ -5,20 +5,21 @@ from collections.abc import Sequence
 from typing import Any, Final
 
 from fastapi import Request
-from pydantic import BaseModel, Field
+from pydantic import Field
 
 from litellm._logging import verbose_proxy_logger
+from litellm.types.llms.base import LiteLLMBaseModel
 
 TrustedProxyNetwork = ipaddress.IPv4Network | ipaddress.IPv6Network
 
 
-class NetworkContext(BaseModel):
+class NetworkContext(LiteLLMBaseModel):
     client_ip: str | None = None
     host: str | None = None
     via_trusted_proxy: bool = False
 
 
-class TrustedProxyConfig(BaseModel):
+class TrustedProxyConfig(LiteLLMBaseModel):
     use_forwarded_for: bool = False
     trusted_proxy_cidrs: Sequence[str] = Field(default_factory=tuple)
 
diff --git a/litellm/proxy/auth/resolvers/grants.py b/litellm/proxy/auth/resolvers/grants.py
index cbfe21924ec..c66761a8459 100644
--- a/litellm/proxy/auth/resolvers/grants.py
+++ b/litellm/proxy/auth/resolvers/grants.py
@@ -14,7 +14,7 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Final, NoReturn, Protocol, TypeAlias
 
 from fastapi import HTTPException, status
-from pydantic import BaseModel, ValidationError
+from pydantic import ValidationError
 from pydantic.main import IncEx
 from typing_extensions import assert_never
 
@@ -32,6 +32,7 @@ from litellm.proxy.auth.auth_checks import (
     get_team_object,
     get_user_object,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.auth.auth_checks import UserNotFoundError
 
 if TYPE_CHECKING:
@@ -133,7 +134,7 @@ GrantOutcome: TypeAlias = ResolvedGrants | GrantDenial | LookupDegraded
 _MODELS_COLUMN: Final[Mapping[str, IncEx | bool]] = MappingProxyType({"models": True})
 
 
-class _UserModelColumn(BaseModel):
+class _UserModelColumn(LiteLLMBaseModel):
     """``LiteLLM_UserTable.models`` is a bare ``list``; re-read it with the shape a token's ``models`` takes."""
 
     models: tuple[str, ...] = ()
diff --git a/litellm/proxy/auth/resolvers/models.py b/litellm/proxy/auth/resolvers/models.py
index 97e66b8fe67..6119bbee6d6 100644
--- a/litellm/proxy/auth/resolvers/models.py
+++ b/litellm/proxy/auth/resolvers/models.py
@@ -2,12 +2,13 @@ from __future__ import annotations
 
 from enum import Enum
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
 
 from litellm.proxy._types import UserAPIKeyAuth
 from litellm.proxy.auth.auth_method import AuthMethod
 from litellm.proxy.auth.network import NetworkContext
 from litellm.proxy.auth.roles import Role, TeamRole
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
 class PrincipalType(str, Enum):
@@ -15,7 +16,7 @@ class PrincipalType(str, Enum):
     SERVICE_ACCOUNT = "service_account"
 
 
-class UserIdentity(BaseModel):
+class UserIdentity(LiteLLMBaseModel):
     id: str
     external_id: str | None = None
     user_name: str | None = None
@@ -23,32 +24,32 @@ class UserIdentity(BaseModel):
     display_name: str | None = None
 
 
-class OrganizationIdentity(BaseModel):
+class OrganizationIdentity(LiteLLMBaseModel):
     id: str
     name: str | None = None
 
 
-class TeamIdentity(BaseModel):
+class TeamIdentity(LiteLLMBaseModel):
     id: str
     name: str | None = None
     role: TeamRole = TeamRole.MEMBER
 
 
-class ProjectIdentity(BaseModel):
+class ProjectIdentity(LiteLLMBaseModel):
     id: str
     name: str | None = None
 
 
-class EndUserIdentity(BaseModel):
+class EndUserIdentity(LiteLLMBaseModel):
     id: str
 
 
-class CredentialRef(BaseModel):
+class CredentialRef(LiteLLMBaseModel):
     key_id: str | None = None
     token_id: str | None = None
 
 
-class Principal(BaseModel):
+class Principal(LiteLLMBaseModel):
     """Normalized caller identity, resolved once per request at the auth seam.
 
     Frozen and constructed fresh per request, never cached or shared. The identity
diff --git a/litellm/proxy/auth/team_grants.py b/litellm/proxy/auth/team_grants.py
index 2029ee342ae..450a3ce456e 100644
--- a/litellm/proxy/auth/team_grants.py
+++ b/litellm/proxy/auth/team_grants.py
@@ -10,7 +10,7 @@ from collections.abc import Mapping, Sequence
 from types import MappingProxyType
 from typing import Annotated, Final
 
-from pydantic import BaseModel, BeforeValidator, ConfigDict, TypeAdapter, ValidationError
+from pydantic import BeforeValidator, ConfigDict, TypeAdapter, ValidationError
 from pydantic.main import IncEx
 from typing_extensions import ReadOnly, TypedDict
 
@@ -20,6 +20,7 @@ from litellm.proxy._types import (
     LiteLLM_TeamTable,
     Member,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 _MODEL_ALIASES_ADAPTER: Final = TypeAdapter(dict[str, str])
 _JSON_COLUMNS: Final[Mapping[str, IncEx | bool]] = MappingProxyType(
@@ -37,13 +38,13 @@ def _decode_model_aliases(value: object) -> object:
         return None
 
 
-class TeamModelAliasTable(BaseModel):
+class TeamModelAliasTable(LiteLLMBaseModel):
     model_config = ConfigDict(protected_namespaces=())
 
     model_aliases: Annotated[Mapping[str, str] | None, BeforeValidator(_decode_model_aliases)] = None
 
 
-class _TeamJsonColumns(BaseModel):
+class _TeamJsonColumns(LiteLLMBaseModel):
     """The two loosely typed columns on ``LiteLLM_TeamTable``, re-read with the shape the badge needs."""
 
     metadata: Mapping[str, object] | None = None
diff --git a/litellm/proxy/batches_endpoints/litellm_executed_batches.py b/litellm/proxy/batches_endpoints/litellm_executed_batches.py
index 7de170880ca..984aed71b1c 100644
--- a/litellm/proxy/batches_endpoints/litellm_executed_batches.py
+++ b/litellm/proxy/batches_endpoints/litellm_executed_batches.py
@@ -32,6 +32,7 @@ from litellm.proxy.openai_files_endpoints.common_utils import LITELLM_EXECUTED_B
 from litellm.proxy.openai_files_endpoints.storage_backend_service import StorageBackendFileService
 from litellm.proxy.utils import PrismaClient, ProxyLogging
 from litellm.repositories.managed_batch_repository import ManagedBatchRepository
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import LiteLLMBatchCreateRequest, OpenAIFileObject, OpenAIFilesPurpose
 from litellm.types.utils import LITELLM_EXECUTED_BATCH_PROVIDERS, ExtractedFileData, LiteLLMBatch, LlmProviders
 
@@ -103,7 +104,7 @@ class _ResultLine(TypedDict):
     error: ReadOnly[_LineError | None]
 
 
-class BatchInputLine(BaseModel):
+class BatchInputLine(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     custom_id: str
diff --git a/litellm/proxy/client/cli/commands/agents.py b/litellm/proxy/client/cli/commands/agents.py
index 28f74c7465e..579d020f936 100644
--- a/litellm/proxy/client/cli/commands/agents.py
+++ b/litellm/proxy/client/cli/commands/agents.py
@@ -13,7 +13,7 @@ from typing import Final, TypeAlias
 
 import click
 import requests
-from pydantic import BaseModel, TypeAdapter, ValidationError
+from pydantic import TypeAdapter, ValidationError
 
 from litellm.proxy.common_utils.codex_model_catalog import (
     CODEX_BASE_INSTRUCTIONS_PATH,
@@ -22,6 +22,7 @@ from litellm.proxy.common_utils.codex_model_catalog import (
     CodexStockModel,
     codex_models_response_json,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .auth import CliContextObj, context_secret_vault, get_stored_api_key, login
 from .claude_settings import ClaudeSettingsError, install_statusline_script
@@ -261,7 +262,7 @@ def agent_launch_args(command: str, base_url: str) -> list[str]:
     return builder(base_url) if builder else []
 
 
-class ListedModel(BaseModel):
+class ListedModel(LiteLLMBaseModel):
     """The fields of a /v1/models entry that an OpenCode or Codex model entry is built from."""
 
     id: str
@@ -270,7 +271,7 @@ class ListedModel(BaseModel):
     max_output_tokens: int | None = None
 
 
-class _ModelListing(BaseModel):
+class _ModelListing(LiteLLMBaseModel):
     data: tuple[ListedModel, ...]
 
 
@@ -318,29 +319,29 @@ def _fetch_model_listing(
     return listing.data
 
 
-class _OpenCodeLimit(BaseModel):
+class _OpenCodeLimit(LiteLLMBaseModel):
     context: int
     output: int
 
 
-class _OpenCodeModel(BaseModel):
+class _OpenCodeModel(LiteLLMBaseModel):
     name: str
     limit: _OpenCodeLimit | None = None
 
 
-class _OpenCodeProviderOptions(BaseModel):
+class _OpenCodeProviderOptions(LiteLLMBaseModel):
     baseURL: str
     apiKey: str
 
 
-class _OpenCodeProvider(BaseModel):
+class _OpenCodeProvider(LiteLLMBaseModel):
     npm: str
     name: str
     options: _OpenCodeProviderOptions
     models: Mapping[str, _OpenCodeModel]
 
 
-class _OpenCodeConfig(BaseModel):
+class _OpenCodeConfig(LiteLLMBaseModel):
     provider: Mapping[str, _OpenCodeProvider]
 
 
diff --git a/litellm/proxy/client/cli/commands/autoroute/config.py b/litellm/proxy/client/cli/commands/autoroute/config.py
index f8738af221e..28a57f71d08 100644
--- a/litellm/proxy/client/cli/commands/autoroute/config.py
+++ b/litellm/proxy/client/cli/commands/autoroute/config.py
@@ -1,7 +1,9 @@
 from collections.abc import Mapping
 from typing import Final, Literal
 
-from pydantic import BaseModel, ConfigDict, Field, JsonValue, TypeAdapter
+from pydantic import ConfigDict, Field, JsonValue, TypeAdapter
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 TIER_NAMES: Final[tuple[str, ...]] = ("SIMPLE", "MEDIUM", "COMPLEX", "REASONING")
 AUTOROUTER_MODEL_NAME: Final = "autorouter"
@@ -11,14 +13,14 @@ class ConfigGenerationError(Exception):
     """Raised when an AutorouteConfig references a model the discovery step didn't find."""
 
 
-class DiscoveredModel(BaseModel):
+class DiscoveredModel(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     name: str
     mode: str = "chat"
 
 
-class _RawModelListing(BaseModel):
+class _RawModelListing(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     id: str
@@ -45,12 +47,12 @@ def embedding_models(models: tuple[DiscoveredModel, ...]) -> tuple[DiscoveredMod
     return tuple(m for m in models if m.mode == "embedding")
 
 
-class HeuristicClassifier(BaseModel):
+class HeuristicClassifier(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     kind: Literal["heuristic"] = "heuristic"
 
 
-class LLMClassifier(BaseModel):
+class LLMClassifier(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     kind: Literal["llm"] = "llm"
     model: str
@@ -60,12 +62,12 @@ class LLMClassifier(BaseModel):
 ClassifierChoice = HeuristicClassifier | LLMClassifier
 
 
-class NoSemanticMatching(BaseModel):
+class NoSemanticMatching(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     kind: Literal["none"] = "none"
 
 
-class KeywordTierRule(BaseModel):
+class KeywordTierRule(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     keywords: tuple[str, ...]
     tier: str
@@ -81,7 +83,7 @@ DEFAULT_KEYWORD_TIER_RULES: Final[tuple[KeywordTierRule, ...]] = (
 )
 
 
-class SemanticMatching(BaseModel):
+class SemanticMatching(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     kind: Literal["semantic"] = "semantic"
     embedding_model: str
@@ -92,7 +94,7 @@ class SemanticMatching(BaseModel):
 SemanticMatchingChoice = NoSemanticMatching | SemanticMatching
 
 
-class AutorouteConfig(BaseModel):
+class AutorouteConfig(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     base_url: str
diff --git a/litellm/proxy/client/cli/commands/claude_settings.py b/litellm/proxy/client/cli/commands/claude_settings.py
index 0c8b2ef9dcf..cdceb91712b 100644
--- a/litellm/proxy/client/cli/commands/claude_settings.py
+++ b/litellm/proxy/client/cli/commands/claude_settings.py
@@ -25,7 +25,7 @@ from typing import Final, TypeAlias
 import click
 from filelock import FileLock
 from packaging.version import InvalidVersion, Version
-from pydantic import BaseModel, ConfigDict, JsonValue, TypeAdapter, ValidationError
+from pydantic import ConfigDict, JsonValue, TypeAdapter, ValidationError
 
 from litellm._version import version as litellm_version
 from litellm.litellm_core_utils.private_json import (
@@ -35,6 +35,7 @@ from litellm.litellm_core_utils.private_json import (
     stage_private_json,
     write_private_bytes,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from . import statusline_script
 from .cmd_quoting import quote_for_cmd
@@ -159,7 +160,7 @@ class StartOn:
 ModelChoice: TypeAlias = KeepModel | UnpinModel | StartOn
 
 
-class OwnedValue(BaseModel):
+class OwnedValue(LiteLLMBaseModel):
     """What one key held at a moment in time; `present=False` is an absent key, not a null one."""
 
     model_config = ConfigDict(frozen=True)
@@ -168,7 +169,7 @@ class OwnedValue(BaseModel):
     value: JsonValue = None
 
 
-class ConfigureReceipt(BaseModel):
+class ConfigureReceipt(LiteLLMBaseModel):
     """What `lite configure claude` found and what it owns, keyed by dotted path (`env.X` or a top-level key).
 
     Ownership moves only by a write: `written` fingerprints the keys some configure changed, at the
diff --git a/litellm/proxy/client/cli/commands/codex_settings.py b/litellm/proxy/client/cli/commands/codex_settings.py
index 5b01c31683a..ca133d3f1e5 100644
--- a/litellm/proxy/client/cli/commands/codex_settings.py
+++ b/litellm/proxy/client/cli/commands/codex_settings.py
@@ -10,7 +10,7 @@ from types import MappingProxyType
 from typing import Final, Literal, TypeAlias
 
 import tomlkit
-from pydantic import BaseModel, ConfigDict, ValidationError
+from pydantic import ConfigDict, ValidationError
 from tomlkit.container import OutOfOrderTableProxy
 from tomlkit.exceptions import TOMLKitError
 from tomlkit.items import InlineTable, Table
@@ -23,6 +23,7 @@ from litellm.litellm_core_utils.private_json import (
     stage_private_bytes,
     stage_private_json,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .agents import CODEX_PROXY_PROVIDER, codex_proxy_provider
 
@@ -37,7 +38,7 @@ class CodexSettingsError(Exception):
     pass
 
 
-class _Receipt(BaseModel):
+class _Receipt(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="forbid")
 
     version: Literal[1] = 1
diff --git a/litellm/proxy/client/cli/commands/configure.py b/litellm/proxy/client/cli/commands/configure.py
index ea24a644084..45c79ccf014 100644
--- a/litellm/proxy/client/cli/commands/configure.py
+++ b/litellm/proxy/client/cli/commands/configure.py
@@ -6,7 +6,8 @@ from typing import Final
 
 import click
 from InquirerPy import inquirer
-from pydantic import BaseModel
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .auth import CliContextObj
 from .claude_settings import (
@@ -38,7 +39,7 @@ from .configure_setup import (
 )
 
 
-class _ConnectionOptions(BaseModel):
+class _ConnectionOptions(LiteLLMBaseModel):
     api_key: str | None = None
     gateway_url: str | None = None
 
@@ -253,7 +254,7 @@ def unconfigure_group(ctx: click.Context, forget: bool) -> None:
             _disconnect(target, forget)
 
 
-class _UnconfigureOptions(BaseModel):
+class _UnconfigureOptions(LiteLLMBaseModel):
     forget: bool = False
 
 
diff --git a/litellm/proxy/client/cli/commands/configure_profiles.py b/litellm/proxy/client/cli/commands/configure_profiles.py
index 8d84dfc2146..3ae651ff6db 100644
--- a/litellm/proxy/client/cli/commands/configure_profiles.py
+++ b/litellm/proxy/client/cli/commands/configure_profiles.py
@@ -9,7 +9,7 @@ from typing import Final, Literal, TypeAlias
 
 import click
 from filelock import FileLock, Timeout
-from pydantic import BaseModel, ConfigDict, Field, ValidationError, field_validator
+from pydantic import ConfigDict, Field, ValidationError, field_validator
 
 from litellm.litellm_core_utils.private_json import (
     commit_staged_json,
@@ -17,6 +17,7 @@ from litellm.litellm_core_utils.private_json import (
     ensure_private_dir,
     stage_private_json,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .agents import codex_config_path
 from .claude_settings import claude_settings_path, configure_state_path
@@ -27,7 +28,7 @@ Target: TypeAlias = Literal["claude", "codex"]
 TARGETS: Final[tuple[Target, ...]] = ("claude", "codex")
 
 
-class SavedSetup(BaseModel):
+class SavedSetup(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="forbid", strict=True)
 
     version: Literal[1] = 1
diff --git a/litellm/proxy/client/cli/commands/configure_setup.py b/litellm/proxy/client/cli/commands/configure_setup.py
index b988b7e95d2..7a262650cc2 100644
--- a/litellm/proxy/client/cli/commands/configure_setup.py
+++ b/litellm/proxy/client/cli/commands/configure_setup.py
@@ -12,13 +12,14 @@ import click
 import requests
 from InquirerPy import inquirer
 from InquirerPy.base.control import Choice
-from pydantic import BaseModel, TypeAdapter, ValidationError
+from pydantic import TypeAdapter, ValidationError
 
 from litellm.proxy.common_utils.model_listing_utils import (
     CLAUDE_CODE_CLIENT,
     CLAUDE_CODE_PICKER_PATTERN,
     GATEWAY_CLIENT_HEADER,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .agents import codex_config_path
 from .claude_settings import (
@@ -68,7 +69,7 @@ _TARGET_SELECTION: Final = TypeAdapter(tuple[Target, ...])
 _MODEL_SELECTION: Final = TypeAdapter(str)
 
 
-class ConnectionSettings(BaseModel):
+class ConnectionSettings(LiteLLMBaseModel):
     base_url: str
     base_url_explicit: bool = False
     api_key: str | None = None
diff --git a/litellm/proxy/client/cli/commands/debug.py b/litellm/proxy/client/cli/commands/debug.py
index 2aca60ca810..3db3662b021 100644
--- a/litellm/proxy/client/cli/commands/debug.py
+++ b/litellm/proxy/client/cli/commands/debug.py
@@ -18,7 +18,9 @@ from typing import Final
 
 import click
 import requests
-from pydantic import BaseModel, ConfigDict, Field, JsonValue, TypeAdapter, ValidationError, field_validator
+from pydantic import ConfigDict, Field, JsonValue, TypeAdapter, ValidationError, field_validator
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from ...http_client import HTTPClient
 from ._cli_context import cli_context_values
@@ -44,7 +46,7 @@ class DebugFailure:
     message: str
 
 
-class ErrorInformation(BaseModel):
+class ErrorInformation(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     error_code: str | None = None
@@ -53,14 +55,14 @@ class ErrorInformation(BaseModel):
     llm_provider: str | None = None
 
 
-class SpendLogMetadata(BaseModel):
+class SpendLogMetadata(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     status: str | None = None
     error_information: ErrorInformation | None = None
 
 
-class SpendLogRow(BaseModel):
+class SpendLogRow(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore", populate_by_name=True)
 
     request_id: str
@@ -96,7 +98,7 @@ class SpendLogRow(BaseModel):
         return self.metadata.error_information
 
 
-class SessionLogsPage(BaseModel):
+class SessionLogsPage(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     data: tuple[SpendLogRow, ...]
@@ -104,7 +106,7 @@ class SessionLogsPage(BaseModel):
     total_pages: int
 
 
-class RequestResponsePayload(BaseModel):
+class RequestResponsePayload(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     proxy_server_request: JsonValue = None
diff --git a/litellm/proxy/client/cli/commands/pi.py b/litellm/proxy/client/cli/commands/pi.py
index 5966a11485d..22c9755ab8e 100644
--- a/litellm/proxy/client/cli/commands/pi.py
+++ b/litellm/proxy/client/cli/commands/pi.py
@@ -16,9 +16,11 @@ from types import MappingProxyType
 from typing import Annotated, Final
 
 import requests
-from pydantic import BaseModel, ConfigDict, JsonValue, TypeAdapter, ValidationError, model_validator
+from pydantic import ConfigDict, JsonValue, TypeAdapter, ValidationError, model_validator
 from pydantic.types import StringConstraints
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 PI_CONFIG_DIR_ENV: Final = "PI_CODING_AGENT_DIR"
 PI_PROVIDER_NAME: Final = "litellm"
 LITELLM_PROXY_API_KEY_ENV: Final = "LITELLM_PROXY_API_KEY"
@@ -55,14 +57,14 @@ class ModelLimits:
 _NonEmptyString = Annotated[str, StringConstraints(min_length=1)]
 
 
-class ListedModel(BaseModel):
+class ListedModel(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     id: _NonEmptyString
     source_model: _NonEmptyString | None = None
 
 
-class _ModelList(BaseModel):
+class _ModelList(LiteLLMBaseModel):
     data: tuple[ListedModel, ...]
 
     @model_validator(mode="after")
@@ -73,13 +75,13 @@ class _ModelList(BaseModel):
         return self
 
 
-class _ModelGroup(BaseModel):
+class _ModelGroup(LiteLLMBaseModel):
     model_group: str
     max_input_tokens: float | None = None
     max_output_tokens: float | None = None
 
 
-class _ModelGroupList(BaseModel):
+class _ModelGroupList(LiteLLMBaseModel):
     data: tuple[_ModelGroup, ...]
 
 
diff --git a/litellm/proxy/client/cli/commands/pkce_login.py b/litellm/proxy/client/cli/commands/pkce_login.py
index 93f5cfea21b..7cf846bdf9a 100644
--- a/litellm/proxy/client/cli/commands/pkce_login.py
+++ b/litellm/proxy/client/cli/commands/pkce_login.py
@@ -20,10 +20,11 @@ from typing import TYPE_CHECKING, Final, Literal, Protocol
 from urllib.parse import parse_qs, urlencode, urlparse
 
 import requests
-from pydantic import BaseModel, ConfigDict, Field, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, TypeAdapter, ValidationError
 from typing_extensions import ReadOnly, TypedDict
 
 from litellm.litellm_core_utils.cli_token_utils import CLI_TOKEN_FRESHNESS_BUFFER_SECONDS
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from .auth import CliTokenData
@@ -35,7 +36,7 @@ _HTTP_TIMEOUT_SECONDS: Final = 15
 _CLIENT_NAME: Final = "litellm-cli"
 
 
-class CliAuthContract(BaseModel):
+class CliAuthContract(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     contract_version: Literal[1]
@@ -48,13 +49,13 @@ class CliAuthContract(BaseModel):
     code_challenge_methods_supported: tuple[str, ...]
 
 
-class _RegisteredClient(BaseModel):
+class _RegisteredClient(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     client_id: str = Field(min_length=1)
 
 
-class _TokenResponse(BaseModel):
+class _TokenResponse(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     access_token: str = Field(min_length=1)
diff --git a/litellm/proxy/common_utils/cache_aware_routing.py b/litellm/proxy/common_utils/cache_aware_routing.py
index 436482a3421..d0685218c87 100644
--- a/litellm/proxy/common_utils/cache_aware_routing.py
+++ b/litellm/proxy/common_utils/cache_aware_routing.py
@@ -8,7 +8,7 @@ from itertools import chain
 from types import MappingProxyType
 from typing import TYPE_CHECKING, Final
 
-from pydantic import BaseModel, ConfigDict, JsonValue, TypeAdapter, ValidationError
+from pydantic import ConfigDict, JsonValue, TypeAdapter, ValidationError
 
 from litellm._logging import verbose_router_logger
 from litellm.caching.dual_cache import DualCache
@@ -17,6 +17,7 @@ from litellm.llms.anthropic.cache_aware_routing import AnthropicCacheRouting, To
 from litellm.proxy._types import UserAPIKeyAuth
 from litellm.proxy.auth.auth_checks import can_key_call_resolved_model
 from litellm.router_strategy.complexity_router.config import ComplexityRouterConfig
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.router import Deployment, PreRoutingHookResponse
 from litellm.types.utils import StandardLoggingRoutingDecision
 
@@ -45,14 +46,14 @@ _CLASSIFIED_CAUSES: Final = frozenset(
 )
 
 
-class _ProxyRequest(BaseModel):
+class _ProxyRequest(LiteLLMBaseModel):
     model_config = ConfigDict(strict=True)
     url: str
     body: Mapping[str, JsonValue]
     headers: Mapping[str, str]
 
 
-class _CallerSettings(BaseModel):
+class _CallerSettings(LiteLLMBaseModel):
     config: Mapping[str, object] | None = None
 
 
diff --git a/litellm/proxy/common_utils/codex_model_catalog.py b/litellm/proxy/common_utils/codex_model_catalog.py
index 177ce6ae9c5..115b31930dc 100644
--- a/litellm/proxy/common_utils/codex_model_catalog.py
+++ b/litellm/proxy/common_utils/codex_model_catalog.py
@@ -26,9 +26,10 @@ from pathlib import Path
 from types import MappingProxyType, NoneType
 from typing import TYPE_CHECKING, Annotated, Final, Literal
 
-from pydantic import BaseModel, ConfigDict, StringConstraints, TypeAdapter, ValidationError
+from pydantic import ConfigDict, StringConstraints, TypeAdapter, ValidationError
 
 from litellm._logging import verbose_proxy_logger
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from litellm.router import Router
@@ -44,7 +45,7 @@ _BODY_PREFIX: Final = '{"models":['
 _BODY_SUFFIX: Final = "]}"
 
 
-class CodexServiceTier(BaseModel):
+class CodexServiceTier(LiteLLMBaseModel):
     """A `ModelServiceTier` as Codex reads it: the slash command is `name` lowercased, and toggling
     it sends `id` as the request's `service_tier`."""
 
@@ -55,7 +56,7 @@ class CodexServiceTier(BaseModel):
     description: str
 
 
-class _ConfiguredTier(BaseModel):
+class _ConfiguredTier(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="forbid")
 
     id: _TierId
@@ -66,12 +67,12 @@ class _ConfiguredTier(BaseModel):
 _CONFIGURED_TIERS: Final = TypeAdapter(tuple[_TierId | _ConfiguredTier, ...])
 
 
-class CodexTruncationPolicy(BaseModel):
+class CodexTruncationPolicy(LiteLLMBaseModel):
     mode: Literal["bytes"] = "bytes"
     limit: int = 10_000
 
 
-class CodexFallbackModel(BaseModel):
+class CodexFallbackModel(LiteLLMBaseModel):
     """One `ModelInfo` entry of a Codex model catalog for a model Codex does not know.
 
     Every field that some Codex release since `model_catalog_json` appeared
@@ -103,13 +104,13 @@ class CodexFallbackModel(BaseModel):
     base_instructions: str
 
 
-class CodexStockUpgrade(BaseModel):
+class CodexStockUpgrade(LiteLLMBaseModel):
     model_config = ConfigDict(extra="allow")
 
     model: str
 
 
-class CodexStockModel(BaseModel):
+class CodexStockModel(LiteLLMBaseModel):
     """One `ModelInfo` entry as Codex ships it or prints it from `codex debug models`.
 
     Only the fields the listing rewrites are named; everything else that release
@@ -129,7 +130,7 @@ class CodexStockModel(BaseModel):
     default_service_tier: str | None = None
 
 
-class CodexStockCatalog(BaseModel):
+class CodexStockCatalog(LiteLLMBaseModel):
     models: tuple[CodexStockModel, ...]
 
 
diff --git a/litellm/proxy/common_utils/periodic_reload_schedule.py b/litellm/proxy/common_utils/periodic_reload_schedule.py
index c3228ee8ac0..4bead482905 100644
--- a/litellm/proxy/common_utils/periodic_reload_schedule.py
+++ b/litellm/proxy/common_utils/periodic_reload_schedule.py
@@ -25,12 +25,13 @@ from typing import (
     cast,  # noqa: TID251  # prisma table access is untyped (PrismaWrapper.__getattr__)
 )
 
-from pydantic import BaseModel, ConfigDict, ValidationError
+from pydantic import ConfigDict, ValidationError
 
 from litellm._logging import verbose_proxy_logger
 from litellm.litellm_core_utils.safe_json_dumps import safe_dumps
 from litellm.proxy.utils import PrismaClient, evict_config_param
 from litellm.repositories.config_repository import ConfigRepository
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from prisma.models import LiteLLM_Config
@@ -80,7 +81,7 @@ class ReloadScheduleStatus(TypedDict):
     next_run: str | None
 
 
-class _IntervalConfig(BaseModel):
+class _IntervalConfig(LiteLLMBaseModel):
     model_config = ConfigDict(strict=True)
 
     interval_hours: int | None = None
diff --git a/litellm/proxy/common_utils/responses_stream_errors.py b/litellm/proxy/common_utils/responses_stream_errors.py
index 25b83a5ba34..723cd88aad0 100644
--- a/litellm/proxy/common_utils/responses_stream_errors.py
+++ b/litellm/proxy/common_utils/responses_stream_errors.py
@@ -9,10 +9,11 @@ from pydantic import BaseModel, ConfigDict, field_validator
 from litellm._logging import redact_internal_details_from_client_message
 from litellm._uuid import uuid
 from litellm.exceptions import MidStreamFallbackError
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import ResponseFailedEvent, ResponsesAPIResponse, ResponsesAPIStreamEvents
 
 
-class _ResponseIdentity(BaseModel):
+class _ResponseIdentity(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, from_attributes=True)
 
     id: str | None = None
@@ -20,7 +21,7 @@ class _ResponseIdentity(BaseModel):
     created_at: int | None = None
 
 
-class _StreamEvent(BaseModel):
+class _StreamEvent(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, from_attributes=True)
 
     type: str | None = None
@@ -28,7 +29,7 @@ class _StreamEvent(BaseModel):
     response: _ResponseIdentity | None = None
 
 
-class _FailureDetails(BaseModel):
+class _FailureDetails(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, from_attributes=True)
 
     message: str | None = None
diff --git a/litellm/proxy/common_utils/semantic_text_index.py b/litellm/proxy/common_utils/semantic_text_index.py
index 030958706cf..7450ef2b509 100644
--- a/litellm/proxy/common_utils/semantic_text_index.py
+++ b/litellm/proxy/common_utils/semantic_text_index.py
@@ -11,9 +11,10 @@ from typing import TYPE_CHECKING, Final, Protocol, TypeAlias
 
 from fastapi import HTTPException
 from openai import OpenAIError
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 
 from litellm.exceptions import BudgetExceededError
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from litellm.proxy._types import UserAPIKeyAuth
@@ -35,19 +36,19 @@ class EmbeddingFailed:
     reason: str
 
 
-class _EmbeddingItem(BaseModel):
+class _EmbeddingItem(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     embedding: tuple[float, ...]
 
 
-class _EmbeddingData(BaseModel):
+class _EmbeddingData(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     data: tuple[_EmbeddingItem, ...]
 
 
-class _EmbeddingRequest(BaseModel):
+class _EmbeddingRequest(LiteLLMBaseModel):
     """The /embeddings-shaped request as the pre-call hooks (rate limits, budgets, guardrails) hand it back."""
 
     model_config = ConfigDict(frozen=True, extra="ignore")
diff --git a/litellm/proxy/common_utils/swagger_utils.py b/litellm/proxy/common_utils/swagger_utils.py
index 83480bf161d..c2a53a37212 100644
--- a/litellm/proxy/common_utils/swagger_utils.py
+++ b/litellm/proxy/common_utils/swagger_utils.py
@@ -1,12 +1,13 @@
 import inspect
 from typing import Any, Final
 
-from pydantic import BaseModel, Field
+from pydantic import Field
 
 from litellm.exceptions import LITELLM_EXCEPTION_TYPES
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class ErrorResponse(BaseModel):
+class ErrorResponse(LiteLLMBaseModel):
     detail: dict[str, Any] = Field(
         ...,
         example={
diff --git a/litellm/proxy/common_utils/timezone_utils.py b/litellm/proxy/common_utils/timezone_utils.py
index 99e89210e43..f7907675fb0 100644
--- a/litellm/proxy/common_utils/timezone_utils.py
+++ b/litellm/proxy/common_utils/timezone_utils.py
@@ -1,12 +1,13 @@
 from datetime import datetime, time, timezone
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 
 import litellm
 from litellm.litellm_core_utils.duration_parser import get_next_standardized_reset_time
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class BudgetResetSettings(BaseModel):
+class BudgetResetSettings(LiteLLMBaseModel):
     """Immutable, validated settings that govern when budgets reset.
 
     Parsed once from `litellm_settings` and injected into consumers (the reset
diff --git a/litellm/proxy/db/baseline_accounting.py b/litellm/proxy/db/baseline_accounting.py
index 625f5d03125..b21b7c8a2b9 100644
--- a/litellm/proxy/db/baseline_accounting.py
+++ b/litellm/proxy/db/baseline_accounting.py
@@ -35,19 +35,20 @@ from litellm.proxy.spend_tracking.baseline_accounting import (
     advance_baseline_history,
 )
 from litellm.proxy.spend_tracking.savings import BaselineCosts, BaselineCostSnapshot, price_baseline_comparison
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from litellm.proxy.utils import PrismaClient
 
 
-class DailyBaselineTarget(BaseModel):
+class DailyBaselineTarget(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True, strict=True)
 
     entity: DailySpendEntity
     entity_id: str | None
 
 
-class DailyBaselineAttribution(BaseModel):
+class DailyBaselineAttribution(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True, strict=True)
 
     date: str
@@ -77,7 +78,7 @@ class DailyBaselineAttribution(BaseModel):
         )
 
 
-class BaselineAccountingRecord(BaseModel):
+class BaselineAccountingRecord(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True, strict=True)
 
     scope: str = Field(pattern=r"^autorouter-baseline:v3:[a-f0-9]{64}$")
@@ -110,7 +111,7 @@ class BaselineAccountingRecord(BaseModel):
         return self
 
 
-class BaselinePublication(BaseModel):
+class BaselinePublication(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True, strict=True)
 
     version: Literal[3] = 3
@@ -154,7 +155,7 @@ def baseline_publication(
     )
 
 
-class _Comparison(BaseModel):
+class _Comparison(LiteLLMBaseModel):
     revision: int
     published_revision: int
     initial_equivalent: bool
@@ -162,14 +163,14 @@ class _Comparison(BaseModel):
     history: str | None
 
 
-class _StoredRecord(BaseModel):
+class _StoredRecord(LiteLLMBaseModel):
     data: str
     publication: str | None
     conflicted: bool
     started_at: float
 
 
-class _Change(BaseModel):
+class _Change(LiteLLMBaseModel):
     request_id: str
     publication: BaselinePublication
     api_key: str
@@ -618,7 +619,7 @@ class BaselineAccountingStore:
             return "unavailable"
 
 
-class _Scope(BaseModel):
+class _Scope(LiteLLMBaseModel):
     scope: str
 
 
diff --git a/litellm/proxy/db/db_transaction_queue/spend_log_cleanup.py b/litellm/proxy/db/db_transaction_queue/spend_log_cleanup.py
index 4ff6d926a99..82ac7828c39 100644
--- a/litellm/proxy/db/db_transaction_queue/spend_log_cleanup.py
+++ b/litellm/proxy/db/db_transaction_queue/spend_log_cleanup.py
@@ -5,7 +5,7 @@ from dataclasses import dataclass
 from datetime import datetime, timedelta, timezone
 from typing import Final, Literal, TypeAlias
 
-from pydantic import BaseModel, TypeAdapter
+from pydantic import TypeAdapter
 
 from litellm._logging import verbose_proxy_logger
 from litellm.caching import RedisCache
@@ -30,6 +30,7 @@ from litellm.proxy.db.db_transaction_queue.spend_logs_partition_manager import (
     SpendLogsPartitionManager,
 )
 from litellm.proxy.utils import PrismaClient
+from litellm.types.llms.base import LiteLLMBaseModel
 
 StopReason: TypeAlias = Literal["exhausted", "budget_exhausted", "batch_cap_reached", "aborted"]
 
@@ -76,7 +77,7 @@ def _record_run_batch(rows_deleted: int) -> None:
         progress.record_batch(rows_deleted)
 
 
-class _RemainingRow(BaseModel):
+class _RemainingRow(LiteLLMBaseModel):
     """One row of the capped outstanding-rows probe, validated out of prisma's untyped result."""
 
     remaining: int
diff --git a/litellm/proxy/db/health_check_latest.py b/litellm/proxy/db/health_check_latest.py
index a825a23841e..9deb7de4ca5 100644
--- a/litellm/proxy/db/health_check_latest.py
+++ b/litellm/proxy/db/health_check_latest.py
@@ -14,10 +14,11 @@ from collections.abc import Sequence
 from datetime import datetime, timezone
 from typing import TYPE_CHECKING, Final
 
-from pydantic import BaseModel, ConfigDict, JsonValue, TypeAdapter, field_validator
+from pydantic import ConfigDict, JsonValue, TypeAdapter, field_validator
 
 from litellm._logging import verbose_proxy_logger
 from litellm.proxy.db.db_span import db_span
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from litellm.proxy.utils import PrismaClient
@@ -44,7 +45,7 @@ ORDER BY "model_id" ASC, "model_name" ASC, "checked_at" DESC
 """
 
 
-class LatestHealthCheckRow(BaseModel):
+class LatestHealthCheckRow(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, protected_namespaces=())
 
     health_check_id: str
diff --git a/litellm/proxy/db/token_auth.py b/litellm/proxy/db/token_auth.py
index 32c83c4f404..1875158f2b1 100644
--- a/litellm/proxy/db/token_auth.py
+++ b/litellm/proxy/db/token_auth.py
@@ -17,10 +17,10 @@ from dataclasses import dataclass
 from datetime import datetime, timedelta, timezone
 from typing import Final, TypeAlias
 
-from pydantic import BaseModel
 from typing_extensions import assert_never
 
 from litellm._logging import verbose_proxy_logger
+from litellm.types.llms.base import LiteLLMBaseModel
 
 IAM_TOKEN_DB_AUTH_ENV_VAR: Final = "IAM_TOKEN_DB_AUTH"
 AZURE_POSTGRESQL_AUTH_ENV_VAR: Final = "AZURE_POSTGRESQL_AUTH"
@@ -217,7 +217,7 @@ def _parse_rds_token_expiration(token: str) -> datetime | None:
         return None
 
 
-class _EntraAccessTokenClaims(BaseModel):
+class _EntraAccessTokenClaims(LiteLLMBaseModel):
     exp: int
 
 
diff --git a/litellm/proxy/guardrails/guardrail_endpoints.py b/litellm/proxy/guardrails/guardrail_endpoints.py
index 4195f319f16..c43c59f11b8 100644
--- a/litellm/proxy/guardrails/guardrail_endpoints.py
+++ b/litellm/proxy/guardrails/guardrail_endpoints.py
@@ -65,6 +65,7 @@ from litellm.types.guardrails import (
     SupportedGuardrailIntegrations,
     ToolPermissionGuardrailConfigModel,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.guardrails.guardrail_hooks.hide_secrets import (
     HideSecretsGuardrailConfigModel,
 )
@@ -342,7 +343,7 @@ async def list_guardrails_v2(
         raise HTTPException(status_code=500, detail=str(e))
 
 
-class CreateGuardrailRequest(BaseModel):
+class CreateGuardrailRequest(LiteLLMBaseModel):
     guardrail: Guardrail
 
 
@@ -453,7 +454,7 @@ async def create_guardrail(
         raise HTTPException(status_code=500, detail=str(e))
 
 
-class UpdateGuardrailRequest(BaseModel):
+class UpdateGuardrailRequest(LiteLLMBaseModel):
     guardrail: Guardrail
 
 
@@ -662,7 +663,7 @@ async def delete_guardrail(
 GENERIC_GUARDRAIL_API: Final = "generic_guardrail_api"
 
 
-class RegisterGuardrailRequest(BaseModel):
+class RegisterGuardrailRequest(LiteLLMBaseModel):
     """Request body for POST /guardrails/register. Follows Generic Guardrail API config."""
 
     guardrail_name: str
@@ -674,21 +675,21 @@ class RegisterGuardrailRequest(BaseModel):
         return dict(self.litellm_params)
 
 
-class RegisterGuardrailResponse(BaseModel):
+class RegisterGuardrailResponse(LiteLLMBaseModel):
     guardrail_id: str
     guardrail_name: str
     status: str
     submitted_at: datetime | None = None
 
 
-class GuardrailSubmissionSummary(BaseModel):
+class GuardrailSubmissionSummary(LiteLLMBaseModel):
     total: int
     pending_review: int
     active: int
     rejected: int
 
 
-class GuardrailSubmissionItem(BaseModel):
+class GuardrailSubmissionItem(LiteLLMBaseModel):
     guardrail_id: str
     guardrail_name: str
     status: str  # pending_review | active | rejected
@@ -706,7 +707,7 @@ class GuardrailSubmissionItem(BaseModel):
     updated_at: datetime | None = None
 
 
-class ListGuardrailSubmissionsResponse(BaseModel):
+class ListGuardrailSubmissionsResponse(LiteLLMBaseModel):
     submissions: list[GuardrailSubmissionItem]
     summary: GuardrailSubmissionSummary
 
@@ -2028,7 +2029,7 @@ async def get_provider_specific_params():
     return provider_params
 
 
-class TestCustomCodeGuardrailRequest(BaseModel):
+class TestCustomCodeGuardrailRequest(LiteLLMBaseModel):
     """Request model for testing custom code guardrails."""
 
     custom_code: str
@@ -2044,7 +2045,7 @@ class TestCustomCodeGuardrailRequest(BaseModel):
     """Optional mock request_data (model, user_id, team_id, metadata, etc.)."""
 
 
-class TestCustomCodeGuardrailResponse(BaseModel):
+class TestCustomCodeGuardrailResponse(LiteLLMBaseModel):
     """Response model for testing custom code guardrails."""
 
     success: bool
diff --git a/litellm/proxy/guardrails/guardrail_hooks/agent_365/agent_365.py b/litellm/proxy/guardrails/guardrail_hooks/agent_365/agent_365.py
index c683e1f2c5d..ac0a6911184 100644
--- a/litellm/proxy/guardrails/guardrail_hooks/agent_365/agent_365.py
+++ b/litellm/proxy/guardrails/guardrail_hooks/agent_365/agent_365.py
@@ -19,7 +19,7 @@ from typing import TYPE_CHECKING, ClassVar, Final, Literal, NoReturn
 
 import httpx
 from fastapi import HTTPException
-from pydantic import BaseModel, ConfigDict, Field, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, TypeAdapter, ValidationError
 from typing_extensions import ReadOnly, TypedDict
 
 from litellm._logging import verbose_proxy_logger
@@ -35,6 +35,7 @@ from litellm.llms.custom_httpx.http_handler import (
     httpxSpecialProvider,
 )
 from litellm.types.guardrails import GuardrailEventHooks
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.guardrails.guardrail_hooks.agent_365 import (
     AGENT_365_PROD_API_BASE,
     AGENT_365_PROD_RESOURCE_APP_ID,
@@ -108,7 +109,7 @@ class _EvaluateResponse(TypedDict, total=False):
     correlationId: ReadOnly[str]
 
 
-class _ToolReference(BaseModel):
+class _ToolReference(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     name: str
diff --git a/litellm/proxy/guardrails/guardrail_hooks/conduct/conduct.py b/litellm/proxy/guardrails/guardrail_hooks/conduct/conduct.py
index c87f8c016b1..077255017f4 100644
--- a/litellm/proxy/guardrails/guardrail_hooks/conduct/conduct.py
+++ b/litellm/proxy/guardrails/guardrail_hooks/conduct/conduct.py
@@ -12,9 +12,10 @@ from functools import partial
 from types import MappingProxyType
 from typing import TYPE_CHECKING, Final, Literal, Protocol
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 
 from litellm.integrations.custom_guardrail import CustomGuardrail, log_guardrail_information
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import ChatCompletionUserMessage
 from litellm.types.proxy.guardrails.guardrail_hooks.conduct import ConductGuardrailConfigModel
 
@@ -60,7 +61,7 @@ def decision_status(decision: ConductDecision) -> GuardrailStatus:
     return "guardrail_flagged" if decision.verdict in FLAGGED_VERDICTS else "success"
 
 
-class ConductVerdict(BaseModel):
+class ConductVerdict(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     verdict: str
diff --git a/litellm/proxy/guardrails/guardrail_hooks/crowdstrike_aidr/crowdstrike_aidr.py b/litellm/proxy/guardrails/guardrail_hooks/crowdstrike_aidr/crowdstrike_aidr.py
index 578825d971e..15803c9a0cf 100644
--- a/litellm/proxy/guardrails/guardrail_hooks/crowdstrike_aidr/crowdstrike_aidr.py
+++ b/litellm/proxy/guardrails/guardrail_hooks/crowdstrike_aidr/crowdstrike_aidr.py
@@ -5,7 +5,7 @@ from collections.abc import Mapping, Sequence
 from typing import TYPE_CHECKING, Annotated, Final, Literal, NamedTuple, Optional, cast
 
 from fastapi import HTTPException
-from pydantic import BaseModel, ConfigDict, Field, ValidationError
+from pydantic import ConfigDict, Field, ValidationError
 from typing_extensions import override
 
 from litellm._logging import verbose_proxy_logger
@@ -26,6 +26,7 @@ from litellm.proxy.common_utils.callback_utils import (
     add_guardrail_to_applied_guardrails_header,
 )
 from litellm.types.guardrails import GuardrailEventHooks, LitellmParams
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import AllMessageValues, OpenAIChatCompletionToolParam
 from litellm.types.proxy.guardrails.guardrail_hooks.crowdstrike_aidr import (
     CrowdStrikeAIDRGuardrailConfigModelOptionalParams,
@@ -41,18 +42,18 @@ class CrowdStrikeAIDRGuardrailMissingSecrets(Exception):
     """Custom exception for missing CrowdStrike AIDR secrets."""
 
 
-class _TextContentPart(BaseModel):
+class _TextContentPart(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid")
 
     type: Literal["text"] = "text"
     text: str
 
 
-class _ImageUrl(BaseModel):
+class _ImageUrl(LiteLLMBaseModel):
     url: str
 
 
-class _ImageUrlContentPart(BaseModel):
+class _ImageUrlContentPart(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid")
 
     type: Literal["image_url"] = "image_url"
@@ -62,17 +63,17 @@ class _ImageUrlContentPart(BaseModel):
 _ContentPart = Annotated[_TextContentPart | _ImageUrlContentPart, Field(discriminator="type")]
 
 
-class _Message(BaseModel):
+class _Message(LiteLLMBaseModel):
     role: str
     content: str | list[_ContentPart] | None = None
 
 
-class _GuardInput(BaseModel):
+class _GuardInput(LiteLLMBaseModel):
     messages: list[_Message]
     tools: Sequence[OpenAIChatCompletionToolParam] | None = None
 
 
-class _GuardChatCompletionsResult(BaseModel):
+class _GuardChatCompletionsResult(LiteLLMBaseModel):
     guard_output: _GuardInput | None = None
     """Updated structured prompt."""
     blocked: bool | None = None
@@ -83,7 +84,7 @@ class _GuardChatCompletionsResult(BaseModel):
     """Result of the policy analyzing and input prompt."""
 
 
-class _GuardChatCompletionsResponse(BaseModel):
+class _GuardChatCompletionsResponse(LiteLLMBaseModel):
     result: _GuardChatCompletionsResult | None = None
 
 
diff --git a/litellm/proxy/guardrails/guardrail_hooks/llm_as_a_judge/__init__.py b/litellm/proxy/guardrails/guardrail_hooks/llm_as_a_judge/__init__.py
index efec15144b8..5085181639f 100644
--- a/litellm/proxy/guardrails/guardrail_hooks/llm_as_a_judge/__init__.py
+++ b/litellm/proxy/guardrails/guardrail_hooks/llm_as_a_judge/__init__.py
@@ -6,7 +6,7 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Final, Generic, Literal, Optional, TypeVar
 
 from fastapi import HTTPException
-from pydantic import BaseModel, ConfigDict, ValidationError
+from pydantic import ConfigDict, ValidationError
 from typing_extensions import NotRequired, ReadOnly, TypedDict, Unpack
 
 import litellm
@@ -21,6 +21,7 @@ from litellm.litellm_core_utils.llm_judge import (
 )
 from litellm.litellm_core_utils.prompt_templates.common_utils import get_last_user_message
 from litellm.types.guardrails import GuardrailEventHooks, Mode, SupportedGuardrailIntegrations
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import LLM_AS_A_JUDGE_GUARDRAIL_CALL_ORIGIN, GenericGuardrailAPIInputs, GuardrailStatus
 
 if TYPE_CHECKING:
@@ -72,7 +73,7 @@ _JUDGE_CALL_METADATA: Final = MappingProxyType(
 )
 
 
-class _LoggedCallParams(BaseModel):
+class _LoggedCallParams(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     metadata: Mapping[str, object] | None = None
diff --git a/litellm/proxy/guardrails/guardrail_hooks/model_armor/file_scanning.py b/litellm/proxy/guardrails/guardrail_hooks/model_armor/file_scanning.py
index 2938f5a6ed8..366ea7326ae 100644
--- a/litellm/proxy/guardrails/guardrail_hooks/model_armor/file_scanning.py
+++ b/litellm/proxy/guardrails/guardrail_hooks/model_armor/file_scanning.py
@@ -19,9 +19,10 @@ from collections.abc import Sequence
 from dataclasses import dataclass
 from typing import Annotated, Final, Literal
 
-from pydantic import BaseModel, ConfigDict, Field, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, TypeAdapter, ValidationError
 
 from litellm._logging import verbose_proxy_logger
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import AllMessageValues
 
 MODEL_ARMOR_MAX_FILE_SIZE_BYTES: Final = 4 * 1024 * 1024
@@ -72,26 +73,26 @@ class FileScanPlan:
     unscannable_count: int
 
 
-class _FileData(BaseModel):
+class _FileData(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
     file_data: str | None = None
     format: str | None = None
     filename: str | None = None
 
 
-class _FileBlock(BaseModel):
+class _FileBlock(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
     type: Literal["file"]
     file: _FileData
 
 
-class _DocumentSource(BaseModel):
+class _DocumentSource(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
     data: str | None = None
     media_type: str | None = None
 
 
-class _DocumentBlock(BaseModel):
+class _DocumentBlock(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
     type: Literal["document"]
     source: _DocumentSource
diff --git a/litellm/proxy/guardrails/guardrail_hooks/panw_prisma_airs/panw_prisma_airs.py b/litellm/proxy/guardrails/guardrail_hooks/panw_prisma_airs/panw_prisma_airs.py
index d51e7c8b8fb..0607996d13d 100644
--- a/litellm/proxy/guardrails/guardrail_hooks/panw_prisma_airs/panw_prisma_airs.py
+++ b/litellm/proxy/guardrails/guardrail_hooks/panw_prisma_airs/panw_prisma_airs.py
@@ -17,7 +17,7 @@ from urllib.parse import urlparse
 
 import httpx
 from fastapi import HTTPException
-from pydantic import BaseModel, ConfigDict, TypeAdapter, ValidationError, field_validator
+from pydantic import ConfigDict, TypeAdapter, ValidationError, field_validator
 
 from litellm._logging import verbose_proxy_logger
 from litellm._uuid import uuid
@@ -43,6 +43,7 @@ from litellm.proxy.common_utils.callback_utils import (
     add_guardrail_to_applied_guardrails_header,
 )
 from litellm.types.guardrails import GuardrailEventHooks
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import AllMessageValues
 from litellm.types.utils import (
     CallTypes,
@@ -67,7 +68,7 @@ ToolCallLike: TypeAlias = (
 )
 
 
-class _ToolCallFunctionSlice(BaseModel):
+class _ToolCallFunctionSlice(LiteLLMBaseModel):
     model_config = ConfigDict(from_attributes=True, extra="ignore")
 
     name: str | None = None
@@ -90,19 +91,19 @@ class _ToolCallFunctionSlice(BaseModel):
         return json.dumps(value) if isinstance(value, (dict, list)) else str(value)
 
 
-class _ToolCallSlice(BaseModel):
+class _ToolCallSlice(LiteLLMBaseModel):
     model_config = ConfigDict(from_attributes=True, extra="ignore")
 
     function: _ToolCallFunctionSlice | None = None
 
 
-class _ResponsesContentPart(BaseModel):
+class _ResponsesContentPart(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     text: str | None = None
 
 
-class _ResponsesInputItem(BaseModel):
+class _ResponsesInputItem(LiteLLMBaseModel):
     """The slice of a raw Responses ``input`` item that decides which ``texts`` it flattens to."""
 
     model_config = ConfigDict(extra="ignore")
diff --git a/litellm/proxy/guardrails/guardrail_hooks/straiker/__init__.py b/litellm/proxy/guardrails/guardrail_hooks/straiker/__init__.py
index cb037fb7513..13f8212e8d0 100644
--- a/litellm/proxy/guardrails/guardrail_hooks/straiker/__init__.py
+++ b/litellm/proxy/guardrails/guardrail_hooks/straiker/__init__.py
@@ -1,10 +1,11 @@
 from typing import TYPE_CHECKING, Final, Literal
 
-from pydantic import BaseModel, field_validator
+from pydantic import field_validator
 
 import litellm
 from litellm._logging import verbose_proxy_logger
 from litellm.types.guardrails import SupportedGuardrailIntegrations
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .straiker import StraikerGuardrail
 
@@ -12,7 +13,7 @@ if TYPE_CHECKING:
     from litellm.types.guardrails import Guardrail, LitellmParams
 
 
-class _V3Routing(BaseModel):
+class _V3Routing(LiteLLMBaseModel):
     api_version: Literal["v1", "v3"] | None = None
     agent_ref: str | None = None
     client: str | None = None
diff --git a/litellm/proxy/guardrails/guardrail_hooks/typesafe/typesafe.py b/litellm/proxy/guardrails/guardrail_hooks/typesafe/typesafe.py
index 96971f0b570..e62f4b70cb8 100644
--- a/litellm/proxy/guardrails/guardrail_hooks/typesafe/typesafe.py
+++ b/litellm/proxy/guardrails/guardrail_hooks/typesafe/typesafe.py
@@ -16,7 +16,7 @@ from typing import TYPE_CHECKING, Annotated, Final, Literal
 import httpx
 from fastapi import HTTPException
 from httpx import Response as HttpxResponse
-from pydantic import BaseModel, ConfigDict, Field, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, TypeAdapter, ValidationError
 
 from litellm._logging import verbose_proxy_logger
 from litellm.compression.compress import get_protected_indices
@@ -33,6 +33,7 @@ from litellm.llms.custom_httpx.http_handler import (
 from litellm.proxy.guardrails.guardrail_hooks.content_text import content_to_text
 from litellm.secret_managers.main import get_secret_str
 from litellm.types.guardrails import GuardrailEventHooks, Mode
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import GenericGuardrailAPIInputs
 
 if TYPE_CHECKING:
@@ -84,14 +85,14 @@ def _safe_response_text(response: HttpxResponse | None, limit: int = 500) -> str
     return (text or "")[:limit]
 
 
-class _JevNoulAnswer(BaseModel):
+class _JevNoulAnswer(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, allow_inf_nan=False)
 
     type: Literal["noul"]
     noul: Annotated[float, Field(ge=0.0, le=1.0)]
 
 
-class _JevSystemOneResponse(BaseModel):
+class _JevSystemOneResponse(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     answers: Mapping[str, _JevNoulAnswer]
diff --git a/litellm/proxy/guardrails/usage_endpoints.py b/litellm/proxy/guardrails/usage_endpoints.py
index fc237bd55c4..2f466f22ef8 100644
--- a/litellm/proxy/guardrails/usage_endpoints.py
+++ b/litellm/proxy/guardrails/usage_endpoints.py
@@ -28,6 +28,7 @@ from litellm.repositories.table_repositories import (
     SpendLogGuardrailIndexRepository,
     SpendLogsRepository,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from prisma import models as prisma_models
@@ -248,7 +249,7 @@ class _DailyPassBlocked(TypedDict):
     blocked: int
 
 
-class UsageOverviewRow(BaseModel):
+class UsageOverviewRow(LiteLLMBaseModel):
     id: str
     name: str
     type: str
@@ -268,7 +269,7 @@ class UsageOverviewRow(BaseModel):
     )
 
 
-class UsageOverviewResponse(BaseModel):
+class UsageOverviewResponse(LiteLLMBaseModel):
     rows: list[UsageOverviewRow]
     chart: list[UsageChartPoint]  # [{ date, passed, blocked }]
     totalRequests: int
@@ -291,13 +292,13 @@ _EMPTY_OVERVIEW: Final = UsageOverviewResponse(
 )
 
 
-class UsageUnitsDailyPoint(BaseModel):
+class UsageUnitsDailyPoint(LiteLLMBaseModel):
     date: str
     units: Mapping[str, int]
     cost: float | None
 
 
-class UsageDetailResponse(BaseModel):
+class UsageDetailResponse(LiteLLMBaseModel):
     guardrail_id: str
     guardrail_name: str
     type: str
@@ -323,7 +324,7 @@ class UsageDetailResponse(BaseModel):
     untracked_usage_units_by_key: Mapping[str, Mapping[str, int]]
 
 
-class UsageLogEntry(BaseModel):
+class UsageLogEntry(LiteLLMBaseModel):
     id: str
     timestamp: str
     action: str  # blocked | passed | flagged | not_run
@@ -335,7 +336,7 @@ class UsageLogEntry(BaseModel):
     reason: str | None
 
 
-class UsageLogsResponse(BaseModel):
+class UsageLogsResponse(LiteLLMBaseModel):
     logs: list[UsageLogEntry]
     total: int
     page: int
diff --git a/litellm/proxy/hooks/autorouter_baseline_cache.py b/litellm/proxy/hooks/autorouter_baseline_cache.py
index 3d577fa60c3..e3cd6c67aa0 100644
--- a/litellm/proxy/hooks/autorouter_baseline_cache.py
+++ b/litellm/proxy/hooks/autorouter_baseline_cache.py
@@ -11,7 +11,7 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Final
 
 import httpx
-from pydantic import BaseModel, ConfigDict, Field, JsonValue, TypeAdapter
+from pydantic import ConfigDict, Field, JsonValue, TypeAdapter
 
 from litellm._logging import verbose_proxy_logger
 from litellm.constants import INTERNAL_CALL_ORIGIN_METADATA_KEY
@@ -37,6 +37,7 @@ from litellm.proxy.spend_tracking.savings import (
     _effective_model_info,  # pyright: ignore[reportPrivateUsage]  # existing deployment-price owner
     _proxy_llm_router,  # pyright: ignore[reportPrivateUsage]  # existing optional proxy-router owner
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.router import BaselineRouteStamp
 from litellm.types.utils import CallTypes, ModelInfo, Usage
 from litellm.utils import get_prompt_cache_min_tokens
@@ -53,7 +54,7 @@ _COUNT_TIMEOUT: Final = 3.0
 _MAX_COUNTS: Final = 4096
 
 
-class CapturedBaselineObservation(BaseModel):
+class CapturedBaselineObservation(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True, strict=True)
 
     scope: str
@@ -75,14 +76,14 @@ class BaselineCacheContext:
     invalidated: str | None = None
 
 
-class _Metadata(BaseModel):
+class _Metadata(LiteLLMBaseModel):
     model_config = ConfigDict(strict=True, arbitrary_types_allowed=True)
     route: BaselineRouteStamp = Field(alias="_autorouter_baseline_route")
     user_api_key_hash: str = Field(min_length=1)
     session_id: str | None = None
 
 
-class _WireEvent(BaseModel):
+class _WireEvent(LiteLLMBaseModel):
     model_config = ConfigDict(strict=True, arbitrary_types_allowed=True)
     httpx_response: httpx.Response
     api_call_start_time: datetime
@@ -92,7 +93,7 @@ class _WireEvent(BaseModel):
     prompt_cache_response_complete: bool = False
 
 
-class _ResponseUsage(BaseModel):
+class _ResponseUsage(LiteLLMBaseModel):
     model_config = ConfigDict(strict=True, from_attributes=True)
     usage: Usage | None = None
 
diff --git a/litellm/proxy/hooks/batch_enqueued_tokens.py b/litellm/proxy/hooks/batch_enqueued_tokens.py
index 7b4fa6fe625..c4591eb17fe 100644
--- a/litellm/proxy/hooks/batch_enqueued_tokens.py
+++ b/litellm/proxy/hooks/batch_enqueued_tokens.py
@@ -17,13 +17,14 @@ from collections.abc import Awaitable, Callable, Mapping, Sequence
 from dataclasses import dataclass, field
 from typing import TYPE_CHECKING, Annotated, Final, Literal, Protocol, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, Field, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, TypeAdapter, ValidationError
 
 from litellm._internal_context import with_service_target
 from litellm._logging import verbose_proxy_logger
 from litellm.caching.redis_cache import log_redis_failure
 from litellm.constants import BATCH_ENQUEUED_TOKEN_LIMIT_METADATA_KEY, BATCH_ENQUEUED_TOKEN_TTL_SECONDS
 from litellm.proxy._types import UserAPIKeyAuth
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from opentelemetry.trace import Span as _Span
@@ -160,7 +161,7 @@ def canonical_provider_batch_id(batch_id: str) -> str:
     return get_original_file_id(batch_id)
 
 
-class _BatchResponseView(BaseModel):
+class _BatchResponseView(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     id: str
diff --git a/litellm/proxy/hooks/batch_rate_limiter.py b/litellm/proxy/hooks/batch_rate_limiter.py
index 894123b256a..b707e72792c 100644
--- a/litellm/proxy/hooks/batch_rate_limiter.py
+++ b/litellm/proxy/hooks/batch_rate_limiter.py
@@ -24,7 +24,7 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Any, Final, Literal, NoReturn, TypeAlias
 
 from fastapi import HTTPException
-from pydantic import BaseModel, Field, TypeAdapter, ValidationError
+from pydantic import Field, TypeAdapter, ValidationError
 
 import litellm
 from litellm._internal_context import with_service_target
@@ -62,6 +62,7 @@ from litellm.proxy.hooks.parallel_request_limiter_v3 import (
     get_or_create_request_stash,
 )
 from litellm.proxy.hooks.rate_limiter_utils import resolve_llm_provider_for_rate_limit
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from opentelemetry.trace import Span as _Span
@@ -101,7 +102,7 @@ _WINDOW_START_ADAPTER: Final[TypeAdapter[int | float | str | None]] = TypeAdapte
 IncrementAmounts: TypeAlias = dict[Literal["requests", "tokens"], int]
 
 
-class BatchFileUsage(BaseModel):
+class BatchFileUsage(LiteLLMBaseModel):
     """
     Internal model for batch file usage tracking, used for batch rate limiting
     """
diff --git a/litellm/proxy/hooks/prompt_cache_prediction.py b/litellm/proxy/hooks/prompt_cache_prediction.py
index e724d973d95..b4f7322b58b 100644
--- a/litellm/proxy/hooks/prompt_cache_prediction.py
+++ b/litellm/proxy/hooks/prompt_cache_prediction.py
@@ -7,12 +7,13 @@ from datetime import datetime
 from typing import TYPE_CHECKING, Final, Literal
 
 import httpx
-from pydantic import BaseModel, ConfigDict, Field, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, TypeAdapter, ValidationError
 
 from litellm._internal_context import with_service_target
 from litellm.caching.dual_cache import DualCache
 from litellm.integrations.custom_logger import CustomLogger
 from litellm.llms.anthropic.prompt_cache_prediction import PromptPrefix, parse_observed_cache
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import ModelResponse
 
 if TYPE_CHECKING:
@@ -21,7 +22,7 @@ if TYPE_CHECKING:
 _RETENTION_SECONDS: Final = 86_400
 
 
-class CacheObservation(BaseModel):
+class CacheObservation(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True, strict=True)
 
     fingerprint: str = Field(pattern=r"^[0-9a-f]{64}$")
@@ -65,19 +66,19 @@ async def _read_exact(cache: DualCache, scope: str, fingerprint: str) -> CacheOb
     return observation if observation.fingerprint == fingerprint else None
 
 
-class _Metadata(BaseModel):
+class _Metadata(LiteLLMBaseModel):
     model_config = ConfigDict(strict=True)
     user_api_key_hash: str = Field(min_length=1)
 
 
-class _Logged(BaseModel):
+class _Logged(LiteLLMBaseModel):
     model_config = ConfigDict(strict=True)
     status: Literal["success"]
     model_id: str = Field(min_length=1)
     metadata: _Metadata
 
 
-class _Event(BaseModel):
+class _Event(LiteLLMBaseModel):
     model_config = ConfigDict(strict=True, arbitrary_types_allowed=True)
     call_type: Literal["anthropic_messages"]
     custom_llm_provider: Literal["anthropic"]
diff --git a/litellm/proxy/lens/endpoints.py b/litellm/proxy/lens/endpoints.py
index 1ac4d96b712..e0b887b3d91 100644
--- a/litellm/proxy/lens/endpoints.py
+++ b/litellm/proxy/lens/endpoints.py
@@ -9,7 +9,7 @@ from uuid import uuid4
 
 from fastapi import APIRouter, Depends, HTTPException, Query, Request, Response
 from fastapi.security import HTTPAuthorizationCredentials, HTTPBearer
-from pydantic import AwareDatetime, BaseModel, Field
+from pydantic import AwareDatetime, Field
 
 from litellm.litellm_core_utils.secret_redaction import redact_internal_details
 from litellm.proxy._types import LitellmUserRoles, ModelAccessDeniedProxyException, ProxyException, UserAPIKeyAuth
@@ -68,6 +68,7 @@ from litellm.proxy.lens.state import (
     summarized,
 )
 from litellm.proxy.tracing_runtime import provide_storage
+from litellm.types.llms.base import LiteLLMBaseModel
 
 router: Final = APIRouter(prefix="/lens", tags=["Lens"])
 _bearer: Final = HTTPBearer()
@@ -397,7 +398,7 @@ async def update_finding(lens_id: str, finding_id: str, body: FindingUpdate, aut
     )
 
 
-class Preview(BaseModel):
+class Preview(LiteLLMBaseModel):
     as_of: AwareDatetime | None = None
     offset: int = Field(default=0, ge=0)
     selection: ActivitySelection
@@ -423,7 +424,7 @@ async def preview_sample(body: Preview, auth: Auth, storage: StorageDep) -> Samp
     )
 
 
-class WorkerBilling(BaseModel):
+class WorkerBilling(LiteLLMBaseModel):
     analysis_key_id: str = Field(pattern=r"^[a-f0-9]{64}$")
 
 
diff --git a/litellm/proxy/lens/inference.py b/litellm/proxy/lens/inference.py
index 491ed9c472f..576e311baf8 100644
--- a/litellm/proxy/lens/inference.py
+++ b/litellm/proxy/lens/inference.py
@@ -5,7 +5,7 @@ from types import MappingProxyType
 from typing import Final
 
 from fastapi import HTTPException, Request
-from pydantic import BaseModel, ConfigDict, Field, field_validator
+from pydantic import ConfigDict, Field, field_validator
 
 import litellm
 from litellm.exceptions import ContextWindowExceededError, ModelNotMappedError
@@ -18,11 +18,12 @@ from litellm.proxy.lens.models import Job, Lens, ModelRequest, ModelResult, Step
 from litellm.proxy.lens.repository import LensRepository
 from litellm.proxy.lens.state import add_step, current_job, renew_budget, replace_job
 from litellm.types.integrations.anthropic_cache_control_hook import CacheControlMessageInjectionPoint
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import AllMessageValues
 from litellm.types.utils import CostPerToken, ModelResponse
 
 
-class DeploymentParams(BaseModel):
+class DeploymentParams(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
     model: str
     input_cost_per_token: float | None = None
@@ -31,30 +32,30 @@ class DeploymentParams(BaseModel):
     max_completion_tokens: int | None = Field(default=None, gt=0)
 
 
-class ModelCapacity(BaseModel):
+class ModelCapacity(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
     max_input_tokens: int | None = Field(default=None, gt=0)
     max_output_tokens: int | None = Field(default=None, gt=0)
 
 
-class Deployment(BaseModel):
+class Deployment(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
     litellm_params: DeploymentParams
     model_info: ModelCapacity = ModelCapacity()
 
 
-class Message(BaseModel):
+class Message(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
     content: str | None = None
 
 
-class Choice(BaseModel):
+class Choice(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
     message: Message
     finish_reason: str | None = None
 
 
-class Completion(BaseModel):
+class Completion(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
     choices: tuple[Choice, ...] = Field(min_length=1)
 
@@ -67,7 +68,7 @@ _SYSTEM: Final = (
 )
 
 
-class Prices(BaseModel):
+class Prices(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
     input_cost_per_token: float = Field(ge=0)
     output_cost_per_token: float = Field(ge=0)
@@ -331,13 +332,13 @@ async def analyze(
     )
 
 
-class Usage(BaseModel):
+class Usage(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
     prompt_tokens: int | None = None
     completion_tokens: int | None = None
 
 
-class UsageEnvelope(BaseModel):
+class UsageEnvelope(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
     model: str | None = None
     usage: Usage | None = None
diff --git a/litellm/proxy/lens/repository.py b/litellm/proxy/lens/repository.py
index ba626ea8dc3..d50299dded6 100644
--- a/litellm/proxy/lens/repository.py
+++ b/litellm/proxy/lens/repository.py
@@ -5,10 +5,11 @@ from collections.abc import AsyncIterator, Awaitable, Callable
 from types import MappingProxyType
 from typing import Final, Protocol
 
-from pydantic import BaseModel, JsonValue, TypeAdapter
+from pydantic import JsonValue, TypeAdapter
 
 from litellm.proxy.db.prisma_client import PrismaWrapper
 from litellm.proxy.lens.models import Job, Lens, Scope, TraceFindingCount, TraceIdentity, Worker
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
 class Database(Protocol):
@@ -16,7 +17,7 @@ class Database(Protocol):
     def execute_raw(self, query: str, *args: object) -> Awaitable[int]: ...
 
 
-class Row(BaseModel):
+class Row(LiteLLMBaseModel):
     data: JsonValue
 
 
diff --git a/litellm/proxy/management_endpoints/auto_router_endpoints.py b/litellm/proxy/management_endpoints/auto_router_endpoints.py
index e372ccedd80..617a76aaa04 100644
--- a/litellm/proxy/management_endpoints/auto_router_endpoints.py
+++ b/litellm/proxy/management_endpoints/auto_router_endpoints.py
@@ -14,7 +14,7 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Annotated, Final, Protocol
 from uuid import uuid4
 
-from pydantic import BaseModel, ConfigDict, TypeAdapter, field_validator
+from pydantic import ConfigDict, TypeAdapter, field_validator
 
 import litellm
 from litellm._logging import verbose_proxy_logger
@@ -62,6 +62,7 @@ from litellm.router_utils.auto_router_model_naming import (
     classify_strategy_router_model,
     strategy_router_dependencies,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.management_endpoints.auto_router_endpoints import (
     SHADOW_EVAL_TURN_VALVE,
     AutoRouterAvailabilityRequest,
@@ -627,7 +628,7 @@ async def preview_auto_router_routing(
     )
 
 
-class _SessionAggRow(BaseModel):
+class _SessionAggRow(LiteLLMBaseModel):
     """One router's window: session shape from overlapping sessions, money from the selected days."""
 
     router_name: str
@@ -1249,7 +1250,7 @@ def _is_unique_violation(error: Exception) -> bool:
     return isinstance(error, UniqueViolationError)
 
 
-class _AttemptAggRow(BaseModel):
+class _AttemptAggRow(LiteLLMBaseModel):
     grp: str
     turn_count: int
     real_wins: int
@@ -1361,7 +1362,7 @@ WHERE group_id = $1 AND stopped_by IS NULL
 """
 
 
-class _FunnelTotalsRow(BaseModel):
+class _FunnelTotalsRow(LiteLLMBaseModel):
     legs_with_rows: int
     not_sampled: int
     unjudgeable: int
@@ -1369,7 +1370,7 @@ class _FunnelTotalsRow(BaseModel):
     withheld: int
 
 
-class _AttemptCountRow(BaseModel):
+class _AttemptCountRow(LiteLLMBaseModel):
     job_id: str
     attempt_count: int
     spend: float
@@ -1395,7 +1396,7 @@ WHERE group_id IN (
 """
 
 
-class _AttemptTotalsRow(BaseModel):
+class _AttemptTotalsRow(LiteLLMBaseModel):
     judged_count: int
     error_count: int
     judge_spend: float
@@ -1429,7 +1430,7 @@ def _leg_group_id(leg: "_LegRow") -> str:
     return leg.group_id
 
 
-class _LegRow(BaseModel):
+class _LegRow(LiteLLMBaseModel):
     """One LiteLLM_ShadowEvalJob row, validated off the untyped prisma record. A row is
     one target's leg of a job; the legs of a job share group_id and identical config,
     written together by one create_many. The API's job id is the group id, so leg ids
diff --git a/litellm/proxy/management_endpoints/cache_settings_endpoints.py b/litellm/proxy/management_endpoints/cache_settings_endpoints.py
index 40124bd19a4..2bd4b6b47f9 100644
--- a/litellm/proxy/management_endpoints/cache_settings_endpoints.py
+++ b/litellm/proxy/management_endpoints/cache_settings_endpoints.py
@@ -15,7 +15,7 @@ from datetime import datetime, timezone
 from typing import TYPE_CHECKING, Any, Final, Protocol
 
 from fastapi import APIRouter, Depends, Header, HTTPException
-from pydantic import BaseModel, Field, TypeAdapter
+from pydantic import Field, TypeAdapter
 
 from litellm._logging import verbose_proxy_logger
 from litellm._redis import _redis_kwargs_from_environment
@@ -30,6 +30,7 @@ from litellm.proxy._types import (
 from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
 from litellm.proxy.db.exception_handler import call_with_db_reconnect_retry
 from litellm.repositories.table_repositories import CacheConfigRepository
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.management_endpoints import (
     CACHE_SETTINGS_FIELDS,
     REDIS_TYPE_DESCRIPTIONS,
@@ -419,23 +420,23 @@ class CacheSettingsManager:
         CacheSettingsManager._last_cache_params = cache_params.copy()
 
 
-class CacheSettingsResponse(BaseModel):
+class CacheSettingsResponse(LiteLLMBaseModel):
     fields: list[CacheSettingsField] = Field(description="List of all configurable cache settings with metadata")
     current_values: dict[str, object] = Field(description="Current values of cache settings")
     redis_type_descriptions: dict[str, str] = Field(description="Descriptions for each Redis type option")
 
 
-class CacheTestRequest(BaseModel):
+class CacheTestRequest(LiteLLMBaseModel):
     cache_settings: dict[str, object] = Field(description="Cache settings to test connection with")
 
 
-class CacheTestResponse(BaseModel):
+class CacheTestResponse(LiteLLMBaseModel):
     status: str = Field(description="Connection status: 'success' or 'failed'")
     message: str = Field(description="Connection result message")
     error: str | None = Field(default=None, description="Error message if connection failed")
 
 
-class CacheSettingsUpdateRequest(BaseModel):
+class CacheSettingsUpdateRequest(LiteLLMBaseModel):
     cache_settings: dict[str, object] = Field(description="Cache settings to save")
 
 
diff --git a/litellm/proxy/management_endpoints/coordination_redis_endpoints.py b/litellm/proxy/management_endpoints/coordination_redis_endpoints.py
index c59ee92f073..fb8b2544d7d 100644
--- a/litellm/proxy/management_endpoints/coordination_redis_endpoints.py
+++ b/litellm/proxy/management_endpoints/coordination_redis_endpoints.py
@@ -19,7 +19,7 @@ from datetime import datetime, timezone
 from typing import Final
 
 from fastapi import APIRouter, Depends, Header, HTTPException
-from pydantic import BaseModel, Field, TypeAdapter, ValidationError
+from pydantic import Field, TypeAdapter, ValidationError
 
 import litellm
 from litellm._logging import verbose_proxy_logger
@@ -39,6 +39,7 @@ from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
 from litellm.proxy.utils import invalidate_config_param
 from litellm.repositories.config_repository import ConfigRepository
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.management_endpoints import (
     COORDINATION_REDIS_SETTINGS_FIELDS,
     CoordinationRedisSettingsField,
@@ -270,7 +271,7 @@ async def _emit_coordination_redis_audit_log(
     task.add_done_callback(_log_audit_task_exception)
 
 
-class CoordinationRedisSettingsResponse(BaseModel):
+class CoordinationRedisSettingsResponse(LiteLLMBaseModel):
     values: dict[str, object] = Field(description="Current coordination Redis settings, with credentials redacted")
     fields: list[CoordinationRedisSettingsField] = Field(
         description="List of all configurable coordination Redis settings with metadata"
@@ -280,11 +281,11 @@ class CoordinationRedisSettingsResponse(BaseModel):
     )
 
 
-class CoordinationRedisSettingsRequest(BaseModel):
+class CoordinationRedisSettingsRequest(LiteLLMBaseModel):
     settings: dict[str, object] = Field(description="Coordination Redis connection params")
 
 
-class CoordinationRedisTestResponse(BaseModel):
+class CoordinationRedisTestResponse(LiteLLMBaseModel):
     status: str = Field(description="Connection status: 'healthy' or 'unhealthy'")
     error: str | None = Field(default=None, description="Error message if the connection failed")
 
diff --git a/litellm/proxy/management_endpoints/cost_tracking_settings.py b/litellm/proxy/management_endpoints/cost_tracking_settings.py
index f6c767cfbb6..469d112675e 100644
--- a/litellm/proxy/management_endpoints/cost_tracking_settings.py
+++ b/litellm/proxy/management_endpoints/cost_tracking_settings.py
@@ -15,7 +15,6 @@ from dataclasses import dataclass
 from typing import Final
 
 from fastapi import APIRouter, Depends, HTTPException
-from pydantic import BaseModel
 
 import litellm
 from litellm._internal_context import current_billing_time, pinned_billing_time
@@ -29,6 +28,7 @@ from litellm.proxy._types import (
 )
 from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
 from litellm.proxy.management_endpoints.prompt_cache_prediction import router as prompt_cache_prediction_router
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import (
     CostBreakdown,
     CostPerToken,
@@ -496,11 +496,11 @@ async def update_cost_margin_config(
         )
 
 
-class BlockUnpricedModelsRequest(BaseModel):
+class BlockUnpricedModelsRequest(LiteLLMBaseModel):
     enabled: bool
 
 
-class BlockUnpricedModelsResponse(BaseModel):
+class BlockUnpricedModelsResponse(LiteLLMBaseModel):
     enabled: bool
 
 
diff --git a/litellm/proxy/management_endpoints/gateway_request_endpoints.py b/litellm/proxy/management_endpoints/gateway_request_endpoints.py
index 898c801d347..c7ae50df91c 100644
--- a/litellm/proxy/management_endpoints/gateway_request_endpoints.py
+++ b/litellm/proxy/management_endpoints/gateway_request_endpoints.py
@@ -16,11 +16,12 @@ from datetime import datetime, timedelta, timezone
 from typing import Annotated, Final
 
 from fastapi import APIRouter, Depends, HTTPException, Query
-from pydantic import BaseModel, TypeAdapter
+from pydantic import TypeAdapter
 
 from litellm._logging import verbose_proxy_logger
 from litellm.proxy._types import CommonProxyErrors, LitellmUserRoles, UserAPIKeyAuth
 from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.gateway_requests import (
     GatewayRequestActivityResponse,
     GatewayRequestBreakdownEntry,
@@ -44,7 +45,7 @@ _AGGREGATE_SQL: Final = """
 """
 
 
-class _AggregateRow(BaseModel):
+class _AggregateRow(LiteLLMBaseModel):
     """Validates one query_raw row so the handler works with typed values, not Any."""
 
     date: str
diff --git a/litellm/proxy/management_endpoints/key_management_endpoints.py b/litellm/proxy/management_endpoints/key_management_endpoints.py
index 2ee29f894b5..6dd9888d925 100644
--- a/litellm/proxy/management_endpoints/key_management_endpoints.py
+++ b/litellm/proxy/management_endpoints/key_management_endpoints.py
@@ -154,6 +154,7 @@ from litellm.repositories.verification_token_repository import (
 from litellm.router import Router
 from litellm.secret_managers.base_secret_manager import raise_if_unsafe_secret_name
 from litellm.secret_managers.main import get_secret
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.management_endpoints.key_management_endpoints import (
     BulkUpdateKeyRequest,
     BulkUpdateKeyResponse,
@@ -212,7 +213,7 @@ class _KeyUpdateResult(TypedDict):
     data: ReadOnly[Mapping[str, object]]
 
 
-class _StoredKeyRouterSettings(BaseModel):
+class _StoredKeyRouterSettings(LiteLLMBaseModel):
     router_settings: Mapping[str, object] | None = None
 
 
@@ -6491,7 +6492,7 @@ KeyStatus = Literal["active", "expired", "revoked", "deleted"]
 VALID_STATUS_FILTER_VALUES: Final[frozenset[KeyStatus]] = frozenset({"active", "expired", "revoked", "deleted"})
 
 
-class _KeyStatusSource(BaseModel):
+class _KeyStatusSource(LiteLLMBaseModel):
     blocked: bool | None = None
     expires: datetime | None = None
 
diff --git a/litellm/proxy/management_endpoints/management_v1/budgets.py b/litellm/proxy/management_endpoints/management_v1/budgets.py
index f1760b857c5..1fe9c94ac5f 100644
--- a/litellm/proxy/management_endpoints/management_v1/budgets.py
+++ b/litellm/proxy/management_endpoints/management_v1/budgets.py
@@ -7,7 +7,7 @@ from types import MappingProxyType
 from typing import Annotated, Final
 
 from fastapi import APIRouter, Depends, Request
-from pydantic import BaseModel, TypeAdapter
+from pydantic import TypeAdapter
 
 from litellm._logging import verbose_proxy_logger
 from litellm.proxy._types import (
@@ -35,6 +35,7 @@ from litellm.proxy.list_api.list_framework import (
 )
 from litellm.proxy.management_endpoints.management_v1.common import MANAGEMENT_V1_PREFIX
 from litellm.proxy.utils import PrismaClient
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.management_endpoints.management_v1 import (
     ListResponse,
     ProblemDetail,
@@ -45,7 +46,7 @@ router: Final = APIRouter(prefix=MANAGEMENT_V1_PREFIX)
 BUDGET_TABLE: Final = '"LiteLLM_BudgetTable"'
 
 
-class BudgetListItem(BaseModel):
+class BudgetListItem(LiteLLMBaseModel):
     """One budget as the Budgets page reads it, and as it comes back off the table.
 
     Validating the raw row through here is what makes `tpm_limit` / `rpm_limit`
@@ -65,7 +66,7 @@ class BudgetListItem(BaseModel):
     updated_at: datetime
 
 
-class _RowCount(BaseModel):
+class _RowCount(LiteLLMBaseModel):
     count: int
 
 
diff --git a/litellm/proxy/management_endpoints/mcp_connector_import.py b/litellm/proxy/management_endpoints/mcp_connector_import.py
index 8120452393c..e3e04eebca0 100644
--- a/litellm/proxy/management_endpoints/mcp_connector_import.py
+++ b/litellm/proxy/management_endpoints/mcp_connector_import.py
@@ -11,13 +11,14 @@ from collections.abc import Mapping
 from dataclasses import dataclass
 from typing import Final
 
-from pydantic import AliasChoices, BaseModel, ConfigDict, Field, ValidationError
+from pydantic import AliasChoices, ConfigDict, Field, ValidationError
 
 from litellm.proxy._types import MCPApprovalStatus, NewMCPServerRequest
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.mcp import MCPAuth, MCPAuthType, MCPCredentials, MCPTransport
 
 
-class MCPConnectorEntry(BaseModel):
+class MCPConnectorEntry(LiteLLMBaseModel):
     model_config = ConfigDict(populate_by_name=True)
 
     name: str | None = None
@@ -33,7 +34,7 @@ class MCPConnectorEntry(BaseModel):
     description: str | None = None
 
 
-class MCPConnectorImportRequest(BaseModel):
+class MCPConnectorImportRequest(LiteLLMBaseModel):
     model_config = ConfigDict(populate_by_name=True)
 
     mcp_servers: Mapping[str, MCPConnectorEntry] | tuple[MCPConnectorEntry, ...] = Field(
@@ -53,23 +54,23 @@ class ConnectorConversionError:
     error: str
 
 
-class MCPConnectorImportResult(BaseModel):
+class MCPConnectorImportResult(LiteLLMBaseModel):
     name: str
     server_id: str
     alias: str
 
 
-class MCPConnectorImportSkipped(BaseModel):
+class MCPConnectorImportSkipped(LiteLLMBaseModel):
     name: str
     reason: str
 
 
-class MCPConnectorImportFailure(BaseModel):
+class MCPConnectorImportFailure(LiteLLMBaseModel):
     name: str
     error: str
 
 
-class MCPConnectorImportResponse(BaseModel):
+class MCPConnectorImportResponse(LiteLLMBaseModel):
     imported: tuple[MCPConnectorImportResult, ...]
     skipped: tuple[MCPConnectorImportSkipped, ...]
     errors: tuple[MCPConnectorImportFailure, ...]
diff --git a/litellm/proxy/management_endpoints/mcp_management_endpoints.py b/litellm/proxy/management_endpoints/mcp_management_endpoints.py
index e5bea71ac7c..e481f331e66 100644
--- a/litellm/proxy/management_endpoints/mcp_management_endpoints.py
+++ b/litellm/proxy/management_endpoints/mcp_management_endpoints.py
@@ -131,11 +131,13 @@ if MCP_AVAILABLE:
             validate_tool_name,  # pyright: ignore[reportAssignmentType]
         )
     except ImportError:
-        from pydantic import BaseModel
+        from pydantic import Field
 
-        class _ToolNameValidationResult(BaseModel):
+        from litellm.types.llms.base import LiteLLMBaseModel
+
+        class _ToolNameValidationResult(LiteLLMBaseModel):
             is_valid: bool = True
-            warnings: list[str] = []
+            warnings: list[str] = Field(default=[])
 
         def validate_tool_name(name: str) -> _ToolNameValidationResult:
             return _ToolNameValidationResult()
diff --git a/litellm/proxy/management_endpoints/model_insights_endpoints.py b/litellm/proxy/management_endpoints/model_insights_endpoints.py
index 53e8dfcc71e..c0e7aaa3f7d 100644
--- a/litellm/proxy/management_endpoints/model_insights_endpoints.py
+++ b/litellm/proxy/management_endpoints/model_insights_endpoints.py
@@ -5,13 +5,14 @@ from datetime import date, datetime, timedelta, timezone
 from typing import Annotated, Final
 
 from fastapi import APIRouter, Depends, HTTPException, Query
-from pydantic import BaseModel, Field, TypeAdapter
+from pydantic import Field, TypeAdapter
 
 from litellm.constants import MODEL_INSIGHTS_DEFAULT_TASK, MODEL_INSIGHTS_MAX_RANGE_DAYS, MODEL_INSIGHTS_TOP_MODELS
 from litellm.proxy._types import CommonProxyErrors, LitellmUserRoles, UserAPIKeyAuth
 from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
 from litellm.proxy.db.model_insights_tasks import load_model_insight_tasks
 from litellm.repositories.table_repositories import DailyModelUsageRepository
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.model_insights import (
     ModelInsightDailyMetric,
     ModelInsightDailyTotal,
@@ -26,7 +27,7 @@ from litellm.types.model_insights import (
 router: Final = APIRouter()
 
 
-class _Sums(BaseModel):
+class _Sums(LiteLLMBaseModel):
     spend: float = 0.0
     prompt_tokens: int = 0
     completion_tokens: int = 0
@@ -35,7 +36,7 @@ class _Sums(BaseModel):
     failed_requests: int = 0
 
 
-class _GroupedModel(BaseModel):
+class _GroupedModel(LiteLLMBaseModel):
     model_group: str
     model: str
     custom_llm_provider: str
@@ -46,7 +47,7 @@ class _GroupedDaily(_GroupedModel):
     date: str
 
 
-class _GroupedDate(BaseModel):
+class _GroupedDate(LiteLLMBaseModel):
     date: str
     sums: _Sums = Field(alias="_sum")
 
diff --git a/litellm/proxy/management_endpoints/model_management_endpoints.py b/litellm/proxy/management_endpoints/model_management_endpoints.py
index b0653d83f15..42c4a15a4ef 100644
--- a/litellm/proxy/management_endpoints/model_management_endpoints.py
+++ b/litellm/proxy/management_endpoints/model_management_endpoints.py
@@ -136,6 +136,7 @@ from litellm.router_utils.auto_router_model_naming import (
     validate_strategy_router_model_write,
 )
 from litellm.router_utils.auto_router_tuning_baseline import is_mutable_tuned_candidate, tuning_quota_violation
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.bedrock import AwsSessionTag
 from litellm.types.proxy.management_endpoints.model_management_endpoints import (
     AutoRouterClassifierDefaultPromptResponse,
@@ -174,7 +175,7 @@ async def update_team(*args, **kwargs):
     return await _legacy_update_team(*args, **kwargs)
 
 
-class UpdatePublicModelGroupsRequest(BaseModel):
+class UpdatePublicModelGroupsRequest(LiteLLMBaseModel):
     """Request model for updating public model groups"""
 
     model_groups: list[str] = Field(description="List of model group names to make public")
@@ -241,7 +242,7 @@ class _TransactionFactory(Protocol):
     def __call__(self, *, timeout: datetime.timedelta = ...) -> AbstractAsyncContextManager[_TxModelTables]: ...
 
 
-class _ModelTransactionClient(BaseModel):
+class _ModelTransactionClient(LiteLLMBaseModel):
     model_config = ConfigDict(arbitrary_types_allowed=True, from_attributes=True)
 
     tx: _TransactionFactory
@@ -3088,7 +3089,7 @@ def _labeled_tiers_from_query(tier_labels: str | None) -> tuple[tuple[Complexity
     return _validated_labeled_tiers(parsed)
 
 
-class AutoRouterClassifierPromptPreviewRequest(BaseModel):
+class AutoRouterClassifierPromptPreviewRequest(LiteLLMBaseModel):
     """A POST rather than query params: the classification sections are the operator's own text,
     which must not reach access logs through a URL."""
 
diff --git a/litellm/proxy/management_endpoints/policy_endpoints/endpoints.py b/litellm/proxy/management_endpoints/policy_endpoints/endpoints.py
index c3c499eeec1..c3d1020f089 100644
--- a/litellm/proxy/management_endpoints/policy_endpoints/endpoints.py
+++ b/litellm/proxy/management_endpoints/policy_endpoints/endpoints.py
@@ -45,6 +45,7 @@ from litellm.proxy.guardrails.guardrail_registry import GuardrailRegistry
 from litellm.proxy.management_helpers.utils import management_endpoint_wrapper
 from litellm.proxy.policy_engine.policy_registry import get_policy_registry
 from litellm.proxy.policy_engine.policy_resolver import PolicyResolver
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.policy_engine import (
     PolicyGuardrailsResponse,
     PolicyInfoResponse,
@@ -258,7 +259,7 @@ def _request_with_json_body(body: dict) -> Request:
     return Request(scope, receive=receive)
 
 
-class TestPoliciesAndGuardrailsRequest(BaseModel):
+class TestPoliciesAndGuardrailsRequest(LiteLLMBaseModel):
     """Request body for POST /utils/test_policies_and_guardrails."""
 
     policy_names: list[str] | None = Field(default=None, description="Policy names to resolve guardrails from")
@@ -449,7 +450,7 @@ async def validate_policy(
     return result
 
 
-class _LoadedPolicy(BaseModel):
+class _LoadedPolicy(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True, hide_input_in_errors=True)
 
     inherit: str | None = None
@@ -459,7 +460,7 @@ class _LoadedPolicy(BaseModel):
     inheritance_chain: tuple[str, ...] = ()
 
 
-class _LoadedPolicies(BaseModel):
+class _LoadedPolicies(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True, hide_input_in_errors=True)
 
     policies: Mapping[str, _LoadedPolicy] = Field(default_factory=dict)
@@ -681,7 +682,7 @@ async def get_policy_templates(
     return _load_policy_templates_from_local_backup()
 
 
-class EnrichTemplateRequest(BaseModel):
+class EnrichTemplateRequest(LiteLLMBaseModel):
     template_id: str
     parameters: dict
     model: str | None = None
@@ -1101,7 +1102,7 @@ def _build_comparison_blocked_words(
     return result
 
 
-class SuggestTemplatesRequest(BaseModel):
+class SuggestTemplatesRequest(LiteLLMBaseModel):
     attack_examples: list[str] = Field(default_factory=list)
     description: str = Field(default="")
     model: str | None = None
@@ -1144,7 +1145,7 @@ class GuardrailTestResultEntry(TypedDict):
     details: str
 
 
-class TestPolicyTemplateRequest(BaseModel):
+class TestPolicyTemplateRequest(LiteLLMBaseModel):
     guardrail_definitions: list[dict] = Field(description="All guardrailDefinitions from the policy template")
     text: str = Field(description="Test input text to run guardrails against")
 
diff --git a/litellm/proxy/management_endpoints/prompt_cache_prediction.py b/litellm/proxy/management_endpoints/prompt_cache_prediction.py
index 441b05e3773..9103fa09893 100644
--- a/litellm/proxy/management_endpoints/prompt_cache_prediction.py
+++ b/litellm/proxy/management_endpoints/prompt_cache_prediction.py
@@ -3,7 +3,7 @@ from types import MappingProxyType
 from typing import Annotated, Final
 
 from fastapi import APIRouter, Depends, HTTPException, Request
-from pydantic import BaseModel, JsonValue, TypeAdapter
+from pydantic import JsonValue, TypeAdapter
 
 from litellm.llms.anthropic.prompt_cache_prediction import (
     TokenCounter,
@@ -23,6 +23,7 @@ from litellm.proxy.hooks.parallel_request_limiter_v3 import (
     _PROXY_MaxParallelRequestsHandler_v3,  # pyright: ignore[reportPrivateUsage]  # use the configured proxy limiter's shared capacity owner
 )
 from litellm.proxy.litellm_pre_call_utils import LiteLLMProxyRequestSetup
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.management_endpoints.prompt_cache_prediction import (
     CachePredictionArm,
     CachePredictionRequest,
@@ -33,7 +34,7 @@ router: Final = APIRouter()
 _REQUEST_DATA: Final = TypeAdapter(Mapping[str, object])
 
 
-class _CallerSettings(BaseModel):
+class _CallerSettings(LiteLLMBaseModel):
     config: Mapping[str, object] | None = None
 
 
diff --git a/litellm/proxy/management_endpoints/prompt_caching_requests.py b/litellm/proxy/management_endpoints/prompt_caching_requests.py
index ff99a78e407..2f00d263d84 100644
--- a/litellm/proxy/management_endpoints/prompt_caching_requests.py
+++ b/litellm/proxy/management_endpoints/prompt_caching_requests.py
@@ -4,7 +4,7 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Annotated, Final
 
 from fastapi import APIRouter, Depends, HTTPException, Query
-from pydantic import BaseModel, Json, TypeAdapter
+from pydantic import Json, TypeAdapter
 
 from litellm.proxy._types import CommonProxyErrors, UserAPIKeyAuth, user_api_key_has_admin_view
 from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
@@ -18,6 +18,7 @@ from litellm.proxy.spend_tracking.spend_tracking_utils import (
     _query_raw_rows,  # pyright: ignore[reportPrivateUsage]  # existing typed spend-query adapter; rows validated below
 )
 from litellm.types.integrations.anthropic_cache_control_hook import GATEWAY_INJECTED_CACHE_METADATA_KEY
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.management_endpoints.prompt_caching_requests import (
     PromptCachingRequest,
     PromptCachingRequestCursor,
@@ -86,7 +87,7 @@ def prompt_caching_requests_sql(filter: PromptCachingRequestFilter) -> str:
     """
 
 
-class _PromptCachingRow(BaseModel):
+class _PromptCachingRow(LiteLLMBaseModel):
     request_id: str
     start_time: datetime
     end_time: datetime
diff --git a/litellm/proxy/management_endpoints/roi_calculator_endpoints.py b/litellm/proxy/management_endpoints/roi_calculator_endpoints.py
index 4d2f53f2227..2c074dafbe6 100644
--- a/litellm/proxy/management_endpoints/roi_calculator_endpoints.py
+++ b/litellm/proxy/management_endpoints/roi_calculator_endpoints.py
@@ -15,7 +15,7 @@ from apscheduler.schedulers.asyncio import (  # pyright: ignore[reportMissingTyp
     AsyncIOScheduler,
 )
 from fastapi import APIRouter, Depends, FastAPI, HTTPException, Query
-from pydantic import BaseModel, ConfigDict, SecretStr, TypeAdapter, ValidationError
+from pydantic import ConfigDict, SecretStr, TypeAdapter, ValidationError
 from starlette.types import Receive, Scope, Send
 
 from litellm.llms.custom_httpx.http_handler import (
@@ -50,6 +50,7 @@ from litellm.proxy.roi_calculator.sync import (
 )
 from litellm.proxy.roi_calculator.sync_store import SyncStore
 from litellm.repositories.config_repository import ConfigRepository
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.roi_calculator import (
     DEFAULT_PROMPT,
     ROIBranchSpend,
@@ -78,7 +79,7 @@ _SYNC_MANAGER: Final = SyncManager()
 _ROI_TAGS: Final[list[str | Enum]] = ["roi calculator"]  # mutable-ok: FastAPI requires list-valued route tags
 
 
-class _RouterEstimatorParams(BaseModel):
+class _RouterEstimatorParams(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", from_attributes=True)
 
     model: str | None = None
@@ -86,14 +87,14 @@ class _RouterEstimatorParams(BaseModel):
     custom_llm_provider: str | None = None
 
 
-class _RouterEstimatorModelInfo(BaseModel):
+class _RouterEstimatorModelInfo(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", from_attributes=True)
 
     base_model: str | None = None
     mode: str | None = None
 
 
-class _RouterEstimatorDeployment(BaseModel):
+class _RouterEstimatorDeployment(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", from_attributes=True)
 
     model_name: str = ""
@@ -270,11 +271,11 @@ def _completion_caller(settings: ROISettings) -> CompletionCaller:
     return complete
 
 
-class _GatewayModel(BaseModel):
+class _GatewayModel(LiteLLMBaseModel):
     id: str
 
 
-class _GatewayModels(BaseModel):
+class _GatewayModels(LiteLLMBaseModel):
     data: tuple[_GatewayModel, ...]
 
 
diff --git a/litellm/proxy/management_endpoints/router_settings_endpoints.py b/litellm/proxy/management_endpoints/router_settings_endpoints.py
index d8ea96e6a46..3f927906ed8 100644
--- a/litellm/proxy/management_endpoints/router_settings_endpoints.py
+++ b/litellm/proxy/management_endpoints/router_settings_endpoints.py
@@ -13,13 +13,14 @@ from types import MappingProxyType
 from typing import Any, Final, get_args
 
 from fastapi import APIRouter, Depends
-from pydantic import BaseModel, Field
+from pydantic import Field
 
 from litellm._logging import verbose_proxy_logger
 from litellm.proxy._types import UserAPIKeyAuth
 from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
 from litellm.proxy.config_resolvers import FieldSource, SettingsStore, source_for
 from litellm.router import Router
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.management_endpoints import (
     ROUTER_SETTINGS_FIELDS,
     ROUTING_STRATEGY_DESCRIPTIONS,
@@ -29,7 +30,7 @@ from litellm.types.management_endpoints import (
 router: Final = APIRouter()
 
 
-class RouterSettingsResponse(BaseModel):
+class RouterSettingsResponse(LiteLLMBaseModel):
     fields: list[RouterSettingsField] = Field(description="List of all configurable router settings with metadata")
     current_values: dict[str, Any] = Field(description="Current values of router settings")
     routing_strategy_descriptions: dict[str, str] = Field(description="Descriptions for each routing strategy option")
@@ -39,7 +40,7 @@ class RouterSettingsResponse(BaseModel):
     source: dict[str, FieldSource] = Field(description="Source of each current router setting")
 
 
-class RouterFieldsResponse(BaseModel):
+class RouterFieldsResponse(LiteLLMBaseModel):
     fields: list[RouterSettingsField] = Field(
         description="List of all configurable router settings with metadata (without field values)"
     )
diff --git a/litellm/proxy/management_endpoints/router_weights.py b/litellm/proxy/management_endpoints/router_weights.py
index 99b368808c3..314f80ad37a 100644
--- a/litellm/proxy/management_endpoints/router_weights.py
+++ b/litellm/proxy/management_endpoints/router_weights.py
@@ -6,6 +6,7 @@ from fastapi import HTTPException
 from pydantic import BaseModel, BeforeValidator, ValidationError
 
 from litellm.repositories.prisma_protocols import TableActions
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.router_weights import RouterWeights
 
 
@@ -36,11 +37,11 @@ class _Router(Protocol):
         pass
 
 
-class _RouterWeightSettings(BaseModel):
+class _RouterWeightSettings(LiteLLMBaseModel):
     weights: RouterWeights | None = None
 
 
-class _RouterWeightModelInfo(BaseModel):
+class _RouterWeightModelInfo(LiteLLMBaseModel):
     team_id: str | None = None
     db_model: bool | None = None
     team_public_model_name: str | None = None
@@ -52,7 +53,7 @@ def _router_weight_model_info(value: object) -> _RouterWeightModelInfo:
     return _RouterWeightModelInfo.model_validate(value or {}, from_attributes=True)
 
 
-class _RouterWeightDeployment(BaseModel):
+class _RouterWeightDeployment(LiteLLMBaseModel):
     model_name: str
     model_info: Annotated[_RouterWeightModelInfo, BeforeValidator(_router_weight_model_info)]
 
diff --git a/litellm/proxy/management_endpoints/scim/scim_v2.py b/litellm/proxy/management_endpoints/scim/scim_v2.py
index a7ee0170325..12e1a0841f4 100644
--- a/litellm/proxy/management_endpoints/scim/scim_v2.py
+++ b/litellm/proxy/management_endpoints/scim/scim_v2.py
@@ -23,7 +23,7 @@ from fastapi import (
     Request,
     Response,
 )
-from pydantic import BaseModel, TypeAdapter, ValidationError
+from pydantic import TypeAdapter, ValidationError
 from typing_extensions import ReadOnly, TypedDict, assert_never
 
 import litellm
@@ -73,6 +73,7 @@ from litellm.repositories.user_repository import UserRepository
 from litellm.repositories.verification_token_repository import (
     VerificationTokenRepository,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.management_endpoints.scim_v2 import *
 
 if TYPE_CHECKING:
@@ -248,7 +249,7 @@ class ScimUserData(TypedDict):
     roles: list[SCIMMultiValuedAttribute] | None
 
 
-class GroupMemberExtractionResult(BaseModel):
+class GroupMemberExtractionResult(LiteLLMBaseModel):
     """Result of extracting and processing group members.
 
     ``all_member_ids`` is deduped order-preserving; ``existing_member_ids`` is not,
diff --git a/litellm/proxy/management_endpoints/team_endpoints.py b/litellm/proxy/management_endpoints/team_endpoints.py
index 0266b452b61..20005862b1a 100644
--- a/litellm/proxy/management_endpoints/team_endpoints.py
+++ b/litellm/proxy/management_endpoints/team_endpoints.py
@@ -33,7 +33,7 @@ from typing import (
 
 import fastapi
 from fastapi import APIRouter, Depends, Header, HTTPException, Request, status
-from pydantic import BaseModel, JsonValue, TypeAdapter, ValidationError
+from pydantic import BaseModel, Field, JsonValue, TypeAdapter, ValidationError
 from typing_extensions import ReadOnly, TypedDict, assert_never
 
 import litellm
@@ -195,6 +195,7 @@ from litellm.repositories.verification_token_repository import (
     VerificationTokenRepository,
 )
 from litellm.router import Router
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.auth.auth_checks import UserNotFoundError
 from litellm.types.proxy.management_endpoints.common_daily_activity import (
     SpendAnalyticsPaginatedResponse,
@@ -4034,7 +4035,7 @@ async def reset_team_member_spend_fn(
     }
 
 
-class _TeamMetadataView(BaseModel):
+class _TeamMetadataView(LiteLLMBaseModel):
     metadata: Mapping[str, object] | None = None
 
 
@@ -4793,11 +4794,11 @@ async def _hydrate_member_user_details(
     return tuple(hydrate(m) for m in members)
 
 
-class _OrganizationModelsRow(BaseModel):
-    models: list[str] = []  # mutable-ok: pydantic field default
+class _OrganizationModelsRow(LiteLLMBaseModel):
+    models: list[str] = Field(default=[])  # mutable-ok: pydantic field default
 
 
-class _TeamRowWithOrganization(BaseModel):
+class _TeamRowWithOrganization(LiteLLMBaseModel):
     litellm_organization_table: _OrganizationModelsRow | None = None
 
 
diff --git a/litellm/proxy/management_endpoints/tool_management_endpoints.py b/litellm/proxy/management_endpoints/tool_management_endpoints.py
index b480d46f185..003b1499582 100644
--- a/litellm/proxy/management_endpoints/tool_management_endpoints.py
+++ b/litellm/proxy/management_endpoints/tool_management_endpoints.py
@@ -15,7 +15,9 @@ from datetime import datetime, timedelta, timezone
 from typing import TYPE_CHECKING, Annotated, Final, Protocol, TypeAlias, TypeVar, overload
 
 from fastapi import APIRouter, Depends, HTTPException, Query
-from pydantic import BaseModel, Field, TypeAdapter
+from pydantic import Field, TypeAdapter
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from prisma.models import LiteLLM_DailyToolSpend as PrismaDailyToolSpendRow
@@ -216,13 +218,13 @@ def _parse_day_start(value: str | None) -> datetime | None:
         )
 
 
-class _ToolSpendSums(BaseModel):
+class _ToolSpendSums(LiteLLMBaseModel):
     spend: float = 0.0
     total_tokens: int = 0
     request_count: int = 0
 
 
-class _TopToolRow(BaseModel):
+class _TopToolRow(LiteLLMBaseModel):
     tool_name: str
     sums: _ToolSpendSums = Field(alias="_sum")
 
diff --git a/litellm/proxy/management_endpoints/ui_sso.py b/litellm/proxy/management_endpoints/ui_sso.py
index 57f52ef4acf..605952b44f0 100644
--- a/litellm/proxy/management_endpoints/ui_sso.py
+++ b/litellm/proxy/management_endpoints/ui_sso.py
@@ -41,7 +41,7 @@ if TYPE_CHECKING:
 import jwt
 from fastapi import APIRouter, Depends, Header, HTTPException, Request, Response, status
 from fastapi.responses import RedirectResponse
-from pydantic import BaseModel, TypeAdapter, ValidationError
+from pydantic import TypeAdapter, ValidationError
 
 import litellm
 from litellm._internal_context import with_service_target
@@ -139,6 +139,7 @@ from litellm.repositories.table_repositories import SSOConfigRepository
 from litellm.repositories.team_repository import TeamRepository
 from litellm.repositories.user_repository import UserRepository
 from litellm.secret_managers.main import get_secret_bool, get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.management_endpoints.ui_sso import *  # noqa: F403
 from litellm.types.proxy.management_endpoints.ui_sso import (
     DefaultTeamSSOParams,
@@ -218,14 +219,14 @@ def _team_detail_db(repo: TeamRepository) -> "TableActions[_TeamDetailRow]":
 _SSO_TOKEN_CLAIMS_ADAPTER: Final = TypeAdapter(Mapping[str, object])
 
 
-class _TeamRowGrants(BaseModel):
+class _TeamRowGrants(LiteLLMBaseModel):
     team_id: str
     team_alias: str | None = None
     models: tuple[str, ...] = ()
     litellm_model_table: TeamModelAliasTable | None = None
 
 
-class CliSsoTeamDetail(BaseModel):
+class CliSsoTeamDetail(LiteLLMBaseModel):
     """The per-team snapshot cached in the CLI SSO flow and echoed to the CLI on poll."""
 
     team_id: str | None = None
diff --git a/litellm/proxy/management_endpoints/usage_endpoints/endpoints.py b/litellm/proxy/management_endpoints/usage_endpoints/endpoints.py
index d1e92d0c7e4..f03259cb4e9 100644
--- a/litellm/proxy/management_endpoints/usage_endpoints/endpoints.py
+++ b/litellm/proxy/management_endpoints/usage_endpoints/endpoints.py
@@ -8,7 +8,7 @@ from typing import Final, Literal
 
 from fastapi import APIRouter, Depends, Request
 from fastapi.responses import StreamingResponse
-from pydantic import BaseModel, Field
+from pydantic import Field
 
 import litellm
 from litellm.proxy._types import UserAPIKeyAuth
@@ -17,16 +17,17 @@ from litellm.proxy.common_utils.sse_keepalive import (
     SSE_COMMENT_PING,
     wrap_sse_stream_with_keepalive_pings,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 router: Final = APIRouter()
 
 
-class ChatMessage(BaseModel):
+class ChatMessage(LiteLLMBaseModel):
     role: Literal["user", "assistant"]
     content: str
 
 
-class UsageAIChatRequest(BaseModel):
+class UsageAIChatRequest(LiteLLMBaseModel):
     messages: list[ChatMessage] = Field(..., description="Chat messages (user/assistant history)")
     model: str | None = Field(default=None, description="Model to use for AI chat")
 
diff --git a/litellm/proxy/management_endpoints/user_agent_analytics_endpoints.py b/litellm/proxy/management_endpoints/user_agent_analytics_endpoints.py
index 0422c72cdb3..348f5815be6 100644
--- a/litellm/proxy/management_endpoints/user_agent_analytics_endpoints.py
+++ b/litellm/proxy/management_endpoints/user_agent_analytics_endpoints.py
@@ -16,7 +16,9 @@ from datetime import datetime, timedelta
 from typing import TYPE_CHECKING, Final, Protocol, TypeVar, overload
 
 from fastapi import APIRouter, Depends, HTTPException, Query
-from pydantic import BaseModel, TypeAdapter
+from pydantic import TypeAdapter
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from prisma.models import LiteLLM_DailyTagSpend as PrismaDailyTagSpendRow
@@ -42,7 +44,7 @@ MAX_TAGS: Final = 250  # Maximum number of distinct tags to return
 router: Final = APIRouter()
 
 
-class TagActiveUsersResponse(BaseModel):
+class TagActiveUsersResponse(LiteLLMBaseModel):
     """Response for tag active users metrics"""
 
     tag: str
@@ -52,13 +54,13 @@ class TagActiveUsersResponse(BaseModel):
     period_end: str | None = None  # For WAU/MAU, this will be the end of the period
 
 
-class ActiveUsersAnalyticsResponse(BaseModel):
+class ActiveUsersAnalyticsResponse(LiteLLMBaseModel):
     """Response for active users analytics"""
 
     results: list[TagActiveUsersResponse]
 
 
-class TagSummaryMetrics(BaseModel):
+class TagSummaryMetrics(LiteLLMBaseModel):
     """Summary metrics for a tag"""
 
     tag: str
@@ -70,25 +72,25 @@ class TagSummaryMetrics(BaseModel):
     total_spend: float
 
 
-class TagSummaryResponse(BaseModel):
+class TagSummaryResponse(LiteLLMBaseModel):
     """Response for tag summary analytics"""
 
     results: list[TagSummaryMetrics]
 
 
-class DistinctTagResponse(BaseModel):
+class DistinctTagResponse(LiteLLMBaseModel):
     """Response for distinct user agent tags"""
 
     tag: str
 
 
-class DistinctTagsResponse(BaseModel):
+class DistinctTagsResponse(LiteLLMBaseModel):
     """Response for all distinct user agent tags"""
 
     results: list[DistinctTagResponse]
 
 
-class PerUserMetrics(BaseModel):
+class PerUserMetrics(LiteLLMBaseModel):
     """Metrics for individual user"""
 
     user_id: str
@@ -101,7 +103,7 @@ class PerUserMetrics(BaseModel):
     spend: float = 0.0
 
 
-class PerUserAnalyticsResponse(BaseModel):
+class PerUserAnalyticsResponse(LiteLLMBaseModel):
     """Response for per-user analytics"""
 
     results: list[PerUserMetrics]
@@ -111,11 +113,11 @@ class PerUserAnalyticsResponse(BaseModel):
     total_pages: int
 
 
-class _DistinctTagRow(BaseModel):
+class _DistinctTagRow(LiteLLMBaseModel):
     tag: str
 
 
-class _ActiveUsersRow(BaseModel):
+class _ActiveUsersRow(LiteLLMBaseModel):
     tag: str
     active_users: int
     date: str
@@ -123,7 +125,7 @@ class _ActiveUsersRow(BaseModel):
     period_end: str | None = None
 
 
-class _TagSummaryRow(BaseModel):
+class _TagSummaryRow(LiteLLMBaseModel):
     tag: str
     unique_users: int | None = None
     total_requests: float | int | str | None = None
diff --git a/litellm/proxy/management_endpoints/workflow_management_endpoints.py b/litellm/proxy/management_endpoints/workflow_management_endpoints.py
index eeb64ab2773..890a344dfe0 100644
--- a/litellm/proxy/management_endpoints/workflow_management_endpoints.py
+++ b/litellm/proxy/management_endpoints/workflow_management_endpoints.py
@@ -23,8 +23,6 @@ try:
     from prisma.errors import UniqueViolationError
 except ImportError:
     UniqueViolationError = None
-from pydantic import BaseModel
-
 from litellm._logging import verbose_proxy_logger
 from litellm.proxy._types import (
     CommonProxyErrors,
@@ -38,6 +36,7 @@ from litellm.repositories.table_repositories import (
     WorkflowMessageRepository,
     WorkflowRunRepository,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 router: Final = APIRouter()
 
@@ -76,7 +75,7 @@ _EVENT_STATUS_MAP: Final[Mapping[str, str]] = {
 # ---------------------------------------------------------------------------
 
 
-class WorkflowRunCreateRequest(BaseModel):
+class WorkflowRunCreateRequest(LiteLLMBaseModel):
     workflow_type: str
     input: Mapping[str, object] | None = None
     metadata: Mapping[str, object] | None = None
@@ -85,19 +84,19 @@ class WorkflowRunCreateRequest(BaseModel):
 WorkflowRunStatus = Literal["pending", "running", "paused", "completed", "failed"]
 
 
-class WorkflowRunUpdateRequest(BaseModel):
+class WorkflowRunUpdateRequest(LiteLLMBaseModel):
     status: WorkflowRunStatus | None = None
     output: Mapping[str, object] | None = None
     metadata: Mapping[str, object] | None = None
 
 
-class WorkflowEventCreateRequest(BaseModel):
+class WorkflowEventCreateRequest(LiteLLMBaseModel):
     event_type: str
     step_name: str
     data: Mapping[str, object] | None = None
 
 
-class WorkflowMessageCreateRequest(BaseModel):
+class WorkflowMessageCreateRequest(LiteLLMBaseModel):
     role: str
     content: str
     session_id: str | None = None
diff --git a/litellm/proxy/management_helpers/access_group_key_sync.py b/litellm/proxy/management_helpers/access_group_key_sync.py
index b9a28a2ebb3..a5289b89cec 100644
--- a/litellm/proxy/management_helpers/access_group_key_sync.py
+++ b/litellm/proxy/management_helpers/access_group_key_sync.py
@@ -28,8 +28,6 @@ schema.
 from collections.abc import Sequence
 from typing import Final, Protocol
 
-from pydantic import BaseModel
-
 from litellm.proxy._types import (
     LiteLLM_VerificationToken,
     RegenerateKeyRequest,
@@ -40,9 +38,10 @@ from litellm.proxy.auth.auth_checks import (
 )
 from litellm.proxy.db.routing_prisma_wrapper import writer_wrapper
 from litellm.repositories.table_repositories import AccessGroupRepository
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class _MovedGroupRow(BaseModel):
+class _MovedGroupRow(LiteLLMBaseModel):
     access_group_id: str
 
 
diff --git a/litellm/proxy/management_helpers/access_group_model_sync.py b/litellm/proxy/management_helpers/access_group_model_sync.py
index 683f2ea79b9..50813b4a2e6 100644
--- a/litellm/proxy/management_helpers/access_group_model_sync.py
+++ b/litellm/proxy/management_helpers/access_group_model_sync.py
@@ -8,19 +8,18 @@ the arrays alone strands every group on a name nothing serves any more.
 from collections.abc import Sequence
 from typing import Final, Protocol
 
-from pydantic import BaseModel
-
 from litellm.proxy.db.routing_prisma_wrapper import writer_wrapper
 from litellm.proxy.management_helpers.access_group_team_sync import invalidate_access_group_caches
 from litellm.repositories.table_repositories import AccessGroupRepository
 from litellm.router import Router
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class _TouchedGroupRow(BaseModel):
+class _TouchedGroupRow(LiteLLMBaseModel):
     access_group_id: str
 
 
-class _DeploymentCountRow(BaseModel):
+class _DeploymentCountRow(LiteLLMBaseModel):
     deployment_count: int
 
 
diff --git a/litellm/proxy/management_helpers/access_group_team_sync.py b/litellm/proxy/management_helpers/access_group_team_sync.py
index 555481d06a4..fbca95b9169 100644
--- a/litellm/proxy/management_helpers/access_group_team_sync.py
+++ b/litellm/proxy/management_helpers/access_group_team_sync.py
@@ -18,10 +18,11 @@ import asyncio
 from collections.abc import Mapping, Sequence
 from typing import Final, Protocol
 
-from pydantic import BaseModel, TypeAdapter
+from pydantic import TypeAdapter
 
 from litellm.proxy.auth.auth_checks import _delete_cache_access_object
 from litellm.proxy.db.db_span import db_span
+from litellm.types.llms.base import LiteLLMBaseModel
 
 # hashtext collisions only cost two unrelated teams a little serialization, and the
 # lock is never taken by the access-group endpoints as a SELECT ... FOR UPDATE row lock,
@@ -58,11 +59,11 @@ RETURNING access_group_id
 """
 
 
-class _AffectedGroup(BaseModel):
+class _AffectedGroup(LiteLLMBaseModel):
     access_group_id: str
 
 
-class _TeamGroups(BaseModel):
+class _TeamGroups(LiteLLMBaseModel):
     access_group_ids: tuple[str, ...] | None = None
 
 
diff --git a/litellm/proxy/management_helpers/auto_router_availability.py b/litellm/proxy/management_helpers/auto_router_availability.py
index 52cd9f0499b..89baf738d75 100644
--- a/litellm/proxy/management_helpers/auto_router_availability.py
+++ b/litellm/proxy/management_helpers/auto_router_availability.py
@@ -4,7 +4,7 @@ from dataclasses import dataclass
 from types import MappingProxyType
 from typing import Final
 
-from pydantic import BaseModel, Json, TypeAdapter, ValidationError
+from pydantic import Json, TypeAdapter, ValidationError
 
 from litellm.proxy.common_utils.encrypt_decrypt_utils import decrypt_value_helper
 from litellm.router_utils.auto_router_model_naming import (
@@ -19,17 +19,18 @@ from litellm.router_utils.auto_router_tuning_baseline import (
     mutable_tuned_identities,
     tuning_quota_violation,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.management_endpoints.auto_router_endpoints import (
     AutoRouterAllowance,
     AutoRouterAvailabilityResponse,
 )
 
 
-class _CatalogModelInfo(BaseModel):
+class _CatalogModelInfo(LiteLLMBaseModel):
     team_id: str | None = None
 
 
-class _CatalogSource(BaseModel):
+class _CatalogSource(LiteLLMBaseModel):
     model_id: str
     created_by: str | None = None
     litellm_params: Json[dict[str, object]] | dict[str, object]
diff --git a/litellm/proxy/management_helpers/auto_router_permissions.py b/litellm/proxy/management_helpers/auto_router_permissions.py
index 5845194fa9b..f2d0aaaec78 100644
--- a/litellm/proxy/management_helpers/auto_router_permissions.py
+++ b/litellm/proxy/management_helpers/auto_router_permissions.py
@@ -5,7 +5,7 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Final, Literal
 
 from fastapi import HTTPException
-from pydantic import BaseModel, ConfigDict, Field, ValidationError
+from pydantic import ConfigDict, Field, ValidationError
 from typing_extensions import ReadOnly, TypedDict
 
 from litellm.models.organization import LiteLLM_OrganizationTable
@@ -38,6 +38,7 @@ from litellm.router_strategy.complexity_router.config import (
     resolve_complexity_router_config_write,
 )
 from litellm.router_utils.auto_router_model_naming import classify_strategy_router_model, strategy_router_dependencies
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.management_endpoints.auto_router_endpoints import RequestComplexityRouterConfig
 from litellm.types.router import Deployment, updateDeployment
 
@@ -45,14 +46,14 @@ if TYPE_CHECKING:
     from prisma import types as prisma_types
 
 
-class _MemberRouterThinking(BaseModel):
+class _MemberRouterThinking(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     type: Literal["enabled", "disabled", "adaptive"]
     budget_tokens: int | None = Field(default=None, gt=0, le=1_000_000)
 
 
-class _MemberRouterGenerationParams(BaseModel):
+class _MemberRouterGenerationParams(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     reasoning_effort: str | None = None
@@ -69,7 +70,7 @@ class _MemberRouterGenerationParams(BaseModel):
     stop: str | tuple[str, ...] | None = None
 
 
-class _MemberOpenSourceClassifierConfig(BaseModel):
+class _MemberOpenSourceClassifierConfig(LiteLLMBaseModel):
     """Classifier settings a team member may set while the gateway owns the connection."""
 
     model_config = ConfigDict(extra="forbid")
@@ -88,7 +89,7 @@ class _MemberComplexityRouterConfig(RequestComplexityRouterConfig):
     model_config = ConfigDict(extra="forbid", arbitrary_types_allowed=True)
 
 
-class _RouterConfigSource(BaseModel):
+class _RouterConfigSource(LiteLLMBaseModel):
     model: str | None = None
     complexity_router_config: Mapping[str, object] | None = None
 
@@ -270,7 +271,7 @@ async def _load_member_auto_router_dependency_objects(
     return MemberAutoRouterDependencyObjects(membership=membership, organization=organization, project=project)
 
 
-class StoredAutoRouterIdentity(BaseModel):
+class StoredAutoRouterIdentity(LiteLLMBaseModel):
     created_by: str | None = None
     updated_at: datetime | None = None
 
diff --git a/litellm/proxy/management_helpers/bulk_team_member_budgets.py b/litellm/proxy/management_helpers/bulk_team_member_budgets.py
index 6821e0ef413..292cb637687 100644
--- a/litellm/proxy/management_helpers/bulk_team_member_budgets.py
+++ b/litellm/proxy/management_helpers/bulk_team_member_budgets.py
@@ -11,7 +11,7 @@ from datetime import datetime, timedelta
 from types import MappingProxyType
 from typing import TYPE_CHECKING, Final
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 
 from litellm.litellm_core_utils.safe_json_dumps import safe_dumps
 from litellm.proxy._types import (
@@ -40,6 +40,7 @@ from litellm.proxy.management_helpers.bulk_user_deletion import (
 )
 from litellm.proxy.utils import PrismaClient
 from litellm.repositories.team_repository import TeamRepository
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.management_endpoints.team_endpoints import (
     BulkTeamMemberBudgetUpdateRequest,
     TeamMemberBudgetPatch,
@@ -86,7 +87,7 @@ async def _shared_budget_ids(tx: "Prisma", budget_ids: frozenset[str]) -> frozen
     return frozenset(budget_id for budget_id in budget_ids if sum(1 for row in rows if row.budget_id == budget_id) > 1)
 
 
-class _AuditedMemberBudget(BaseModel):
+class _AuditedMemberBudget(LiteLLMBaseModel):
     """One member's limits as the audit log's before/after values record them."""
 
     model_config = ConfigDict(frozen=True)
@@ -101,7 +102,7 @@ class _AuditedMemberBudget(BaseModel):
     allowed_models: tuple[str, ...] | None = None
 
 
-class _AuditedMemberBudgets(BaseModel):
+class _AuditedMemberBudgets(LiteLLMBaseModel):
     """The audit-log columns hold a JSON object, so the per-member list is nested under a key."""
 
     model_config = ConfigDict(frozen=True)
diff --git a/litellm/proxy/management_helpers/bulk_user_creation.py b/litellm/proxy/management_helpers/bulk_user_creation.py
index af62ea62a1a..ecaec1b6260 100644
--- a/litellm/proxy/management_helpers/bulk_user_creation.py
+++ b/litellm/proxy/management_helpers/bulk_user_creation.py
@@ -13,7 +13,7 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Final, Literal, TypeAlias, TypeVar
 
 from fastapi import HTTPException, Request
-from pydantic import BaseModel, ConfigDict, TypeAdapter, ValidationError
+from pydantic import ConfigDict, TypeAdapter, ValidationError
 from typing_extensions import ReadOnly, TypedDict
 
 from litellm._logging import verbose_proxy_logger
@@ -59,6 +59,7 @@ from litellm.proxy.utils import PrismaClient
 from litellm.repositories.prisma_protocols import TableActions
 from litellm.repositories.team_repository import TeamRepository
 from litellm.repositories.user_repository import UserRepository
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.management_endpoints.internal_user_endpoints import (
     BulkNewUserItem,
     BulkNewUserMeta,
@@ -96,7 +97,7 @@ class _PendingUser:
     teams: tuple[NewUserRequestTeam, ...]
 
 
-class _UserRow(BaseModel):
+class _UserRow(LiteLLMBaseModel):
     """The `/user/new` body after defaults and object permission were applied."""
 
     model_config = ConfigDict(extra="ignore")
@@ -176,7 +177,7 @@ _ERROR_DETAIL: Final = TypeAdapter(Mapping[str, object])
 _JSON_OBJECT: Final = TypeAdapter(dict[str, object])
 
 
-class _KeyResponse(BaseModel):
+class _KeyResponse(LiteLLMBaseModel):
     token: str
 
 
diff --git a/litellm/proxy/management_helpers/model_allowlist_rename_sync.py b/litellm/proxy/management_helpers/model_allowlist_rename_sync.py
index f93312f7a37..f1a9faf6b2e 100644
--- a/litellm/proxy/management_helpers/model_allowlist_rename_sync.py
+++ b/litellm/proxy/management_helpers/model_allowlist_rename_sync.py
@@ -11,15 +11,14 @@ from dataclasses import dataclass
 from types import MappingProxyType
 from typing import Final
 
-from pydantic import BaseModel
-
 from litellm.proxy.common_utils.auth_cache_invalidation_pubsub import evict_and_broadcast
 from litellm.proxy.common_utils.user_api_key_cache import UserApiKeyCache
 from litellm.proxy.management_helpers.access_group_model_sync import raw_executor, still_backed
 from litellm.router import Router
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class _TouchedRow(BaseModel):
+class _TouchedRow(LiteLLMBaseModel):
     kind: str
     object_id: str
     team_alias: str | None = None
diff --git a/litellm/proxy/management_helpers/team_metadata_validation.py b/litellm/proxy/management_helpers/team_metadata_validation.py
index 8bb32696857..6171874f491 100644
--- a/litellm/proxy/management_helpers/team_metadata_validation.py
+++ b/litellm/proxy/management_helpers/team_metadata_validation.py
@@ -16,9 +16,10 @@ from types import MappingProxyType
 from typing import Final, Literal, Protocol
 
 from fastapi import HTTPException, status
-from pydantic import BaseModel, JsonValue, TypeAdapter
+from pydantic import JsonValue, TypeAdapter
 
 from litellm.proxy._types import CommonProxyErrors, UserAPIKeyAuth
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.management_endpoints.team_endpoints import (
     TeamMetadataFieldSchema,
 )
@@ -30,13 +31,13 @@ DEFAULT_TEAM_METADATA_VALIDATION_UNAVAILABLE_MESSAGE: Final = (
 DEFAULT_TEAM_METADATA_VALIDATION_REJECTED_MESSAGE: Final = "Team metadata failed validation."
 
 
-class TeamMetadataRequester(BaseModel):
+class TeamMetadataRequester(LiteLLMBaseModel):
     user_id: str | None = None
     user_email: str | None = None
     user_role: str | None = None
 
 
-class TeamMetadataValidationPayload(BaseModel):
+class TeamMetadataValidationPayload(LiteLLMBaseModel):
     operation: Literal["create", "update"]
     metadata: Mapping[str, JsonValue]
     existing_metadata: Mapping[str, JsonValue] | None = None
@@ -45,7 +46,7 @@ class TeamMetadataValidationPayload(BaseModel):
     requester: TeamMetadataRequester
 
 
-class TeamMetadataValidationResult(BaseModel):
+class TeamMetadataValidationResult(LiteLLMBaseModel):
     valid: bool
     error_message: str | None = None
 
diff --git a/litellm/proxy/pass_through_endpoints/llm_provider_handlers/transcribe_passthrough_logging_handler.py b/litellm/proxy/pass_through_endpoints/llm_provider_handlers/transcribe_passthrough_logging_handler.py
index e08a600594e..a8dfb745290 100644
--- a/litellm/proxy/pass_through_endpoints/llm_provider_handlers/transcribe_passthrough_logging_handler.py
+++ b/litellm/proxy/pass_through_endpoints/llm_provider_handlers/transcribe_passthrough_logging_handler.py
@@ -14,7 +14,7 @@ from urllib.parse import quote
 
 import httpx
 import soundfile
-from pydantic import BaseModel, ConfigDict, TypeAdapter, ValidationError
+from pydantic import ConfigDict, TypeAdapter, ValidationError
 from typing_extensions import ReadOnly, TypedDict
 
 import litellm
@@ -45,6 +45,7 @@ from litellm.proxy.common_utils.resource_ownership import (
     is_proxy_admin,
     user_can_access_resource_owner,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.custom_http import httpxSpecialProvider
 from litellm.types.utils import StandardPassThroughResponseObject
 
@@ -72,18 +73,18 @@ class GetTranscriptionJobRequest(TypedDict):
     TranscriptionJobName: ReadOnly[str]
 
 
-class _MediaRef(BaseModel):
+class _MediaRef(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     MediaFileUri: str | None = None
 
 
-class _JobTag(BaseModel):
+class _JobTag(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     Key: str | None = None
     Value: str | None = None
 
 
-class TranscriptionJobRecord(BaseModel):
+class TranscriptionJobRecord(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     TranscriptionJobStatus: str | None = None
     CreationTime: float | None = None
@@ -91,7 +92,7 @@ class TranscriptionJobRecord(BaseModel):
     Tags: tuple[_JobTag, ...] = ()
 
 
-class _TranscriptionJobResponse(BaseModel):
+class _TranscriptionJobResponse(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     TranscriptionJob: TranscriptionJobRecord | None = None
 
@@ -105,7 +106,7 @@ StartedJob: TypeAlias = TranscriptionJobRecord | None
 JobPricer: TypeAlias = Callable[[str, str, float, StartedJob], Awaitable[float]]
 
 
-class _PricedCostMapEntry(BaseModel):
+class _PricedCostMapEntry(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, strict=True)
     input_cost_per_second: float
 
diff --git a/litellm/proxy/pass_through_endpoints/llm_provider_handlers/typesafe_passthrough_logging_handler.py b/litellm/proxy/pass_through_endpoints/llm_provider_handlers/typesafe_passthrough_logging_handler.py
index 0fa2bfff218..9f8318f9b74 100644
--- a/litellm/proxy/pass_through_endpoints/llm_provider_handlers/typesafe_passthrough_logging_handler.py
+++ b/litellm/proxy/pass_through_endpoints/llm_provider_handlers/typesafe_passthrough_logging_handler.py
@@ -3,7 +3,7 @@ from datetime import datetime
 from typing import TYPE_CHECKING, Final
 
 import httpx
-from pydantic import BaseModel, TypeAdapter, ValidationError
+from pydantic import TypeAdapter, ValidationError
 
 import litellm
 from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
@@ -11,23 +11,24 @@ from litellm.litellm_core_utils.litellm_logging import (
     get_standard_logging_object_payload,  # pyright: ignore[reportUnknownVariableType]  # legacy helper has an untyped signature
 )
 from litellm.llms.laya.common_utils import laya_response_model
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import ModelResponse, StandardPassThroughResponseObject, Usage
 
 if TYPE_CHECKING:
     from litellm.proxy._types import PassThroughEndpointLoggingTypedDict
 
 
-class _TypeSafeUsage(BaseModel):
+class _TypeSafeUsage(LiteLLMBaseModel):
     input_tokens: int = 0
     output_tokens: int = 0
 
 
-class _TypeSafeResponse(BaseModel):
+class _TypeSafeResponse(LiteLLMBaseModel):
     model: str | None = None
     usage: _TypeSafeUsage | None = None
 
 
-class _RegistryPricing(BaseModel):
+class _RegistryPricing(LiteLLMBaseModel):
     input_cost_per_token: float = 0.0
     output_cost_per_token: float = 0.0
 
diff --git a/litellm/proxy/prometheus_metrics_server.py b/litellm/proxy/prometheus_metrics_server.py
index 01479f9dd41..053accaf650 100644
--- a/litellm/proxy/prometheus_metrics_server.py
+++ b/litellm/proxy/prometheus_metrics_server.py
@@ -23,11 +23,12 @@ from typing import Final
 import httpx
 from fastapi import FastAPI
 from prometheus_client import CollectorRegistry, multiprocess
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 from starlette.types import ASGIApp, Message, Receive, Scope, Send
 
 from litellm.integrations.prometheus_metrics_endpoint import make_metrics_asgi_app
 from litellm.llms.custom_httpx.http_handler import HTTPHandler
+from litellm.types.llms.base import LiteLLMBaseModel
 
 METRICS_PATH: Final = "/metrics"
 HEALTH_PATH: Final = "/health"
@@ -39,7 +40,7 @@ _STARTUP_PROBE_TIMEOUT_SECONDS: Final = 1.0
 _WILDCARD_TO_LOOPBACK: Final = MappingProxyType({"0.0.0.0": "127.0.0.1", "::": "::1"})
 
 
-class _CliArgs(BaseModel):
+class _CliArgs(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     host: str
diff --git a/litellm/proxy/prompts/prompt_endpoints.py b/litellm/proxy/prompts/prompt_endpoints.py
index 06957fd1c0f..41c8a6a77b4 100644
--- a/litellm/proxy/prompts/prompt_endpoints.py
+++ b/litellm/proxy/prompts/prompt_endpoints.py
@@ -36,6 +36,7 @@ from litellm.proxy.prompts.prompt_registry import (
     prompt_environment_or_default,
 )
 from litellm.repositories.table_repositories import PromptRepository
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.prompts.init_prompts import (
     ListPromptsResponse,
     PromptInfo,
@@ -76,7 +77,7 @@ class _PromptRow(Protocol):
     def model_dump(self) -> Mapping[str, object]: ...
 
 
-class _PromptRowData(BaseModel):
+class _PromptRowData(LiteLLMBaseModel):
     prompt_id: str
     version: int = 1
     environment: str = "development"
@@ -190,7 +191,7 @@ def create_versioned_prompt_spec(db_prompt: _PromptRow) -> PromptSpec:
     )
 
 
-class Prompt(BaseModel):
+class Prompt(LiteLLMBaseModel):
     prompt_id: str
     litellm_params: PromptLiteLLMParams
     prompt_info: PromptInfo | None = None
@@ -211,7 +212,7 @@ def is_ambiguous_keyed_prompt_data(litellm_params: PromptLiteLLMParams) -> bool:
     return bool(prompt_data) and "content" not in prompt_data
 
 
-class PatchPromptRequest(BaseModel):
+class PatchPromptRequest(LiteLLMBaseModel):
     litellm_params: PromptLiteLLMParams | None = None
     prompt_info: PromptInfo | None = None
 
@@ -1039,7 +1040,6 @@ async def test_prompt(
         }'
     ```
     """
-    from pydantic import BaseModel
 
     from litellm.integrations.dotprompt.dotprompt_manager import DotpromptManager
     from litellm.integrations.dotprompt.prompt_manager import (
diff --git a/litellm/proxy/proxy_cli.py b/litellm/proxy/proxy_cli.py
index 08dfc79def7..1a741af6654 100644
--- a/litellm/proxy/proxy_cli.py
+++ b/litellm/proxy/proxy_cli.py
@@ -15,7 +15,7 @@ import click
 import httpx
 from click.core import ParameterSource
 from dotenv import load_dotenv
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 
 import litellm
 from litellm.constants import DEFAULT_NUM_WORKERS_LITELLM_PROXY
@@ -26,6 +26,7 @@ from litellm.proxy.db.pgbouncer import (
     start_in_container_pgbouncer,
 )
 from litellm.proxy.db.query_engine_reaper import start_query_engine_reaper
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from fastapi import FastAPI
@@ -97,7 +98,7 @@ def _build_db_connection_url_params(
     return params
 
 
-class DatabaseTimeoutSettings(BaseModel):
+class DatabaseTimeoutSettings(LiteLLMBaseModel):
     """The `general_settings` keys that bound how long a statement may hold locks.
 
     Validated at the boundary so a mistyped value fails at startup with a clear
diff --git a/litellm/proxy/response_api_endpoints/endpoints.py b/litellm/proxy/response_api_endpoints/endpoints.py
index 882aa240fad..20461cc9a01 100644
--- a/litellm/proxy/response_api_endpoints/endpoints.py
+++ b/litellm/proxy/response_api_endpoints/endpoints.py
@@ -12,7 +12,7 @@ from fastapi import APIRouter, Depends, HTTPException, Request, Response
 from fastapi.responses import JSONResponse
 from openai.types.responses import ResponseItemList
 from openai.types.responses.response_create_params import ResponseInputParam
-from pydantic import BaseModel, ConfigDict, ValidationError
+from pydantic import ConfigDict, ValidationError
 from starlette.websockets import WebSocket, WebSocketDisconnect
 from typing_extensions import ReadOnly, TypedDict
 
@@ -35,6 +35,7 @@ from litellm.proxy.common_utils.http_parsing_utils import (
     _safe_set_request_parsed_body,
 )
 from litellm.proxy.route_llm_request import raise_if_required_body_param_missing
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import (
     REASONING_EFFORT,
     ResponsesAPIOptionalRequestParams,
@@ -1396,7 +1397,7 @@ def _extract_model_from_first_ws_event(first_event: object) -> str | None:
     return (nested.get("model") if isinstance(nested, dict) else None) or first_event.get("model")
 
 
-class _ResponseCreateRoutingHints(BaseModel):
+class _ResponseCreateRoutingHints(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True)
 
     input: str | Sequence[object] | None = None
diff --git a/litellm/proxy/roi_calculator/github.py b/litellm/proxy/roi_calculator/github.py
index 993b6b9c9fa..922faaed992 100644
--- a/litellm/proxy/roi_calculator/github.py
+++ b/litellm/proxy/roi_calculator/github.py
@@ -6,13 +6,14 @@ from typing import Final, TypeVar
 from urllib.parse import quote
 
 import httpx
-from pydantic import BaseModel, ConfigDict, Field, TypeAdapter
+from pydantic import ConfigDict, Field, TypeAdapter
 from typing_extensions import ReadOnly, TypedDict
 
 from litellm.llms.custom_httpx.http_handler import (
     get_async_httpx_client,  # pyright: ignore[reportUnknownVariableType]  # shared client factory has untyped params
 )
 from litellm.proxy.roi_calculator.analytics import normalize_email
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.custom_http import httpxSpecialProvider
 from litellm.types.roi_calculator import ROIPullCommit, ROIPullEvidence, ROIPullFile, ROISettings
 from litellm.types.roi_observed import ObservedIssue
@@ -24,7 +25,7 @@ class SourceError(Exception):
     pass
 
 
-class _GitHubModel(BaseModel):
+class _GitHubModel(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
 
@@ -333,7 +334,7 @@ class _Issue(_GitHubModel):
     pull_request: object | None = None
 
 
-class GitHubIssueSettings(BaseModel):
+class GitHubIssueSettings(LiteLLMBaseModel):
     has_issues: bool
 
 
diff --git a/litellm/proxy/roi_calculator/github_observed.py b/litellm/proxy/roi_calculator/github_observed.py
index a34d34823be..74d3557e203 100644
--- a/litellm/proxy/roi_calculator/github_observed.py
+++ b/litellm/proxy/roi_calculator/github_observed.py
@@ -2,38 +2,39 @@ from datetime import date, datetime, time, timedelta, timezone
 from typing import Final, Literal
 
 import httpx
-from pydantic import BaseModel, Field
+from pydantic import Field
 
 from litellm.proxy.roi_calculator.github import GitHubIssueSettings, GitHubPullListItem, SourceError, request_github
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.roi_calculator import ROISettings
 from litellm.types.roi_observed import ObservedIssue
 
 
-class _PageInfo(BaseModel):
+class _PageInfo(LiteLLMBaseModel):
     hasNextPage: bool = False
     endCursor: str | None = None
 
 
-class _Author(BaseModel):
+class _Author(LiteLLMBaseModel):
     login: str
     kind: str = Field(alias="__typename")
     email: str | None = None
 
 
-class _Repository(BaseModel):
+class _Repository(LiteLLMBaseModel):
     nameWithOwner: str
 
 
-class _Label(BaseModel):
+class _Label(LiteLLMBaseModel):
     name: str
 
 
-class _Labels(BaseModel):
+class _Labels(LiteLLMBaseModel):
     nodes: tuple[_Label, ...] = ()
     pageInfo: _PageInfo = Field(default_factory=_PageInfo)
 
 
-class _Node(BaseModel):
+class _Node(LiteLLMBaseModel):
     number: int
     url: str
     title: str
@@ -69,17 +70,17 @@ class _Node(BaseModel):
         )
 
 
-class _Search(BaseModel):
+class _Search(LiteLLMBaseModel):
     issueCount: int
     pageInfo: _PageInfo
     nodes: tuple[_Node, ...]
 
 
-class _Data(BaseModel):
+class _Data(LiteLLMBaseModel):
     search: _Search
 
 
-class _Response(BaseModel):
+class _Response(LiteLLMBaseModel):
     data: _Data | None = None
     errors: tuple[object, ...] = ()
 
diff --git a/litellm/proxy/roi_calculator/gitlab.py b/litellm/proxy/roi_calculator/gitlab.py
index 9fbf43d8890..ae53583ac5c 100644
--- a/litellm/proxy/roi_calculator/gitlab.py
+++ b/litellm/proxy/roi_calculator/gitlab.py
@@ -14,6 +14,7 @@ from litellm.llms.custom_httpx.http_handler import (
 from litellm.proxy.roi_calculator.analytics import normalize_email
 from litellm.proxy.roi_calculator.github import GitHubPullListItem, SourceError
 from litellm.proxy.roi_calculator.source import repository_tag
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.custom_http import httpxSpecialProvider
 from litellm.types.roi_calculator import ROIPullCommit, ROIPullEvidence, ROIPullFile, ROISettings
 from litellm.types.roi_observed import ObservedIssue
@@ -21,13 +22,13 @@ from litellm.types.roi_observed import ObservedIssue
 _T: Final = TypeVar("_T", bound=BaseModel)
 
 
-class _User(BaseModel):
+class _User(LiteLLMBaseModel):
     username: str
     public_email: str | None = None
     bot: bool = False
 
 
-class _Project(BaseModel):
+class _Project(LiteLLMBaseModel):
     id: int
     path_with_namespace: str
     visibility: str = "private"
@@ -36,7 +37,7 @@ class _Project(BaseModel):
     issues_access_level: str = "enabled"
 
 
-class _MergeRequest(BaseModel):
+class _MergeRequest(LiteLLMBaseModel):
     iid: int
     title: str
     description: str | None = None
@@ -70,7 +71,7 @@ class _MergeRequest(BaseModel):
         )
 
 
-class _Diff(BaseModel):
+class _Diff(LiteLLMBaseModel):
     new_path: str
     old_path: str
     diff: str = ""
@@ -95,12 +96,12 @@ class _Diff(BaseModel):
         )
 
 
-class _Commit(BaseModel):
+class _Commit(LiteLLMBaseModel):
     id: str
     message: str
 
 
-class _Issue(BaseModel):
+class _Issue(LiteLLMBaseModel):
     iid: int
     created_at: datetime
     labels: tuple[str, ...] = ()
diff --git a/litellm/proxy/roi_calculator/oauth.py b/litellm/proxy/roi_calculator/oauth.py
index 32d8cd3e1d0..68202ed5a8d 100644
--- a/litellm/proxy/roi_calculator/oauth.py
+++ b/litellm/proxy/roi_calculator/oauth.py
@@ -13,7 +13,7 @@ from urllib.parse import parse_qsl, urlencode, urlsplit
 import httpx
 from fastapi import HTTPException
 from oauthlib.oauth2 import WebApplicationClient
-from pydantic import BaseModel, ConfigDict, Field, SecretStr, TypeAdapter
+from pydantic import ConfigDict, Field, SecretStr, TypeAdapter
 
 from litellm.llms.custom_httpx.http_handler import (
     get_async_httpx_client,  # pyright: ignore[reportUnknownVariableType]  # shared client factory has untyped params
@@ -30,6 +30,7 @@ from litellm.proxy.roi_calculator.settings import (
 )
 from litellm.proxy.roi_calculator.sync_store import SyncStore
 from litellm.repositories.config_repository import ConfigRepository
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.custom_http import httpxSpecialProvider
 from litellm.types.roi_calculator import ROISettings, ROISyncStatus
 
@@ -111,7 +112,7 @@ def oauth_config(provider: Provider) -> OAuthConfig | None:
     )
 
 
-class OAuthState(BaseModel):
+class OAuthState(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     provider: Provider
@@ -124,11 +125,11 @@ class OAuthState(BaseModel):
     flow: Literal["authorize", "install"] = "authorize"
 
 
-class _Envelope(BaseModel):
+class _Envelope(LiteLLMBaseModel):
     payload: str
 
 
-class _StateRow(BaseModel):
+class _StateRow(LiteLLMBaseModel):
     param_value: _Envelope
 
 
@@ -218,7 +219,7 @@ async def consume_state(
     return value
 
 
-class TokenGrant(BaseModel):
+class TokenGrant(LiteLLMBaseModel):
     access_token: SecretStr
     token_type: str = "bearer"
     refresh_token: SecretStr = SecretStr("")
diff --git a/litellm/proxy/roi_calculator/settings.py b/litellm/proxy/roi_calculator/settings.py
index 89a691843f4..add8285b138 100644
--- a/litellm/proxy/roi_calculator/settings.py
+++ b/litellm/proxy/roi_calculator/settings.py
@@ -5,12 +5,13 @@ from types import MappingProxyType
 from typing import Annotated, Final, Literal
 
 from fastapi import Depends, HTTPException
-from pydantic import BaseModel, ConfigDict, Field, SecretStr, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, SecretStr, TypeAdapter, ValidationError
 
 from litellm.proxy._types import CommonProxyErrors, LitellmUserRoles, UserAPIKeyAuth
 from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
 from litellm.proxy.common_utils.encrypt_decrypt_utils import decrypt_value_helper, encrypt_value_helper
 from litellm.repositories.config_repository import ConfigRepository
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.roi_calculator import DEFAULT_PROMPT, ROISettings
 
 _SETTINGS_KEY: Final = "roi_calculator_settings"
@@ -20,7 +21,7 @@ def connection_id(provider: str, api_url: str) -> str:
     return provider + "_" + sha256(api_url.strip().rstrip("/").encode()).hexdigest()[:16]
 
 
-class StoredConnection(BaseModel):
+class StoredConnection(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     source_provider: Literal["github", "gitlab"]
@@ -38,7 +39,7 @@ class StoredConnection(BaseModel):
         return connection_id(self.source_provider, self.api_url)
 
 
-class StoredROISettings(BaseModel):
+class StoredROISettings(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     revision: int = 0
diff --git a/litellm/proxy/roi_calculator/sync.py b/litellm/proxy/roi_calculator/sync.py
index de9a1a979f1..d134ec8bea0 100644
--- a/litellm/proxy/roi_calculator/sync.py
+++ b/litellm/proxy/roi_calculator/sync.py
@@ -8,7 +8,7 @@ from typing import Final, Literal, NamedTuple, Protocol, runtime_checkable
 from uuid import uuid4
 
 import httpx
-from pydantic import BaseModel, ConfigDict, Field, TypeAdapter
+from pydantic import ConfigDict, Field, TypeAdapter
 from typing_extensions import ReadOnly, TypedDict, Unpack
 
 from litellm._logging import verbose_proxy_logger
@@ -18,6 +18,7 @@ from litellm.proxy.roi_calculator.github import GitHubPullListItem, SourceError
 from litellm.proxy.roi_calculator.pull_cache import cache_key, settings_fingerprint
 from litellm.proxy.roi_calculator.source import RepositorySource, create_source, repository_tag
 from litellm.repositories.chunked_in import find_many_in
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.roi_calculator import (
     ROIBranchSpend,
     ROIEstimate,
@@ -99,12 +100,12 @@ def spend_prisma_client(prisma_client: object) -> _SpendPrismaClient:
     return prisma_client
 
 
-class _DailySpendSums(BaseModel):
+class _DailySpendSums(LiteLLMBaseModel):
     spend: float = 0.0
     api_requests: int = 0
 
 
-class _DailySpendGroup(BaseModel):
+class _DailySpendGroup(LiteLLMBaseModel):
     model_config = ConfigDict(from_attributes=True)
 
     user_id: str | None
@@ -112,7 +113,7 @@ class _DailySpendGroup(BaseModel):
     sums: _DailySpendSums = Field(alias="_sum")
 
 
-class _UserEmail(BaseModel):
+class _UserEmail(LiteLLMBaseModel):
     model_config = ConfigDict(from_attributes=True)
 
     user_id: str
diff --git a/litellm/proxy/roi_calculator/sync_store.py b/litellm/proxy/roi_calculator/sync_store.py
index df126da6d70..c6263d48d43 100644
--- a/litellm/proxy/roi_calculator/sync_store.py
+++ b/litellm/proxy/roi_calculator/sync_store.py
@@ -2,9 +2,10 @@ from datetime import datetime, timezone
 from types import MappingProxyType
 from typing import Final, Literal, Protocol, cast  # noqa: TID251 - PrismaWrapper dynamically delegates database methods
 
-from pydantic import BaseModel, ConfigDict, TypeAdapter
+from pydantic import ConfigDict, TypeAdapter
 
 from litellm.proxy.utils import PrismaClient
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.roi_calculator import ROIReport, ROISyncStatus
 from litellm.types.roi_observed import ObservedData
 
@@ -12,13 +13,13 @@ _SYNC_KEY: Final = "roi_calculator_sync"
 _REPORT_KEY: Final = "roi_calculator_report"
 
 
-class _SyncState(BaseModel):
+class _SyncState(LiteLLMBaseModel):
     owner: str
     status: ROISyncStatus
     cancel: bool = False
 
 
-class _StateRow(BaseModel):
+class _StateRow(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
     param_value: _SyncState
     expired: bool = False
diff --git a/litellm/proxy/search_endpoints/search_tool_management.py b/litellm/proxy/search_endpoints/search_tool_management.py
index 6c06836c190..9d4304e1d47 100644
--- a/litellm/proxy/search_endpoints/search_tool_management.py
+++ b/litellm/proxy/search_endpoints/search_tool_management.py
@@ -7,7 +7,6 @@ from datetime import datetime
 from typing import Any, Final, TypeAlias
 
 from fastapi import APIRouter, Depends, HTTPException
-from pydantic import BaseModel
 
 from litellm._logging import verbose_proxy_logger
 from litellm.constants import UI_SESSION_TOKEN_TEAM_ID
@@ -21,6 +20,7 @@ from litellm.proxy.search_endpoints.search_tool_registry import (
     SearchToolRegistry,
     keep_loaded_search_tools_that_do_not_decrypt,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.search import (
     ListSearchToolsResponse,
     SearchTool,
@@ -277,7 +277,7 @@ async def list_search_tools(
         raise HTTPException(status_code=500, detail=str(e))
 
 
-class CreateSearchToolRequest(BaseModel):
+class CreateSearchToolRequest(LiteLLMBaseModel):
     search_tool: SearchTool
 
 
@@ -349,7 +349,7 @@ async def create_search_tool(request: CreateSearchToolRequest):
         raise HTTPException(status_code=500, detail=str(e))
 
 
-class UpdateSearchToolRequest(BaseModel):
+class UpdateSearchToolRequest(LiteLLMBaseModel):
     search_tool: SearchTool
 
 
@@ -566,7 +566,7 @@ async def get_search_tool_info(search_tool_id: str):
         raise HTTPException(status_code=500, detail=str(e))
 
 
-class TestSearchToolConnectionRequest(BaseModel):
+class TestSearchToolConnectionRequest(LiteLLMBaseModel):
     litellm_params: dict[str, Any]
 
 
diff --git a/litellm/proxy/spend_tracking/baseline_accounting.py b/litellm/proxy/spend_tracking/baseline_accounting.py
index 263dc4de529..46a38f71260 100644
--- a/litellm/proxy/spend_tracking/baseline_accounting.py
+++ b/litellm/proxy/spend_tracking/baseline_accounting.py
@@ -13,16 +13,17 @@ from math import isfinite
 from types import MappingProxyType
 from typing import Final, Literal
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
 
 from litellm.llms.anthropic.prompt_cache_prediction import CountedBreakpoint, CountedPromptCachePlan
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import CacheCreationTokenDetails, PromptTokensDetailsWrapper, Usage
 
 MAX_CACHE_TTL: Final = 3600
 MAX_CACHE_ENTRIES: Final = 1024
 
 
-class BaselineObservation(BaseModel):
+class BaselineObservation(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True, strict=True)
 
     version: Literal[3] = 3
diff --git a/litellm/proxy/spend_tracking/key_metadata_recovery.py b/litellm/proxy/spend_tracking/key_metadata_recovery.py
index 900fea7e2cc..5442c0f3f81 100644
--- a/litellm/proxy/spend_tracking/key_metadata_recovery.py
+++ b/litellm/proxy/spend_tracking/key_metadata_recovery.py
@@ -6,7 +6,7 @@ from datetime import datetime, timedelta
 from types import MappingProxyType
 from typing import Final, Literal, TypeVar
 
-from pydantic import BaseModel, TypeAdapter
+from pydantic import TypeAdapter
 from typing_extensions import ReadOnly, TypedDict
 
 from litellm._logging import verbose_proxy_logger
@@ -24,6 +24,7 @@ from litellm.proxy.db.db_span import db_span, db_spanned
 from litellm.proxy.utils import PrismaClient
 from litellm.repositories.chunked_in import find_many_in
 from litellm.repositories.user_repository import UserRepository
+from litellm.types.llms.base import LiteLLMBaseModel
 
 _T = TypeVar("_T")
 
@@ -118,7 +119,7 @@ class KeyMetadataDict(TypedDict, total=False):
     key_exists: ReadOnly[bool]
 
 
-class _TokenDigestRow(BaseModel):
+class _TokenDigestRow(LiteLLMBaseModel):
     digest: str
     key_alias: str | None = None
     team_id: str | None = None
@@ -131,7 +132,7 @@ def _unanimous(first: str | None, last: str | None) -> str | None:
     return first if last is None or first == last else None
 
 
-class _SpendLogDigestRow(BaseModel):
+class _SpendLogDigestRow(LiteLLMBaseModel):
     digest: str
     first_alias: str | None = None
     last_alias: str | None = None
@@ -148,7 +149,7 @@ class _SpendLogDigestRow(BaseModel):
         )
 
 
-class _DailyUserSpendOwnerRow(BaseModel):
+class _DailyUserSpendOwnerRow(LiteLLMBaseModel):
     api_key: str
     first_owner: str | None = None
     last_owner: str | None = None
diff --git a/litellm/proxy/spend_tracking/savings.py b/litellm/proxy/spend_tracking/savings.py
index fbcf9c78d3e..aa0590ce934 100644
--- a/litellm/proxy/spend_tracking/savings.py
+++ b/litellm/proxy/spend_tracking/savings.py
@@ -14,7 +14,7 @@ from math import isclose, isfinite
 from types import MappingProxyType
 from typing import TYPE_CHECKING, Final, Literal, NamedTuple
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
 
 import litellm
 from litellm._logging import verbose_proxy_logger
@@ -28,6 +28,7 @@ from litellm.types.integrations.anthropic_cache_control_hook import (
     GATEWAY_INJECTED_CACHE_METADATA_KEY,
     GATEWAY_INJECTED_FOR_EVERY_DEPLOYMENT,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from litellm.router import Router
@@ -122,7 +123,7 @@ class PricingBasis(NamedTuple):
 _STANDARD_RATES: Final = PricingBasis()
 
 
-class BaselineCostSnapshot(BaseModel):
+class BaselineCostSnapshot(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True, strict=True)
 
     model: str
diff --git a/litellm/proxy/spend_tracking/spend_capture_rate.py b/litellm/proxy/spend_tracking/spend_capture_rate.py
index 664f100a7b2..443010b65da 100644
--- a/litellm/proxy/spend_tracking/spend_capture_rate.py
+++ b/litellm/proxy/spend_tracking/spend_capture_rate.py
@@ -11,7 +11,7 @@ from datetime import date, datetime, timedelta, timezone
 from types import MappingProxyType
 from typing import TYPE_CHECKING, Final, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, TypeAdapter
+from pydantic import ConfigDict, TypeAdapter
 from typing_extensions import assert_never
 
 from litellm._internal_context import with_service_target
@@ -30,6 +30,7 @@ from litellm.llms.openai.organization_costs import (
 )
 from litellm.proxy.db.db_transaction_queue.pod_lock_manager import POD_LOCK_TARGET
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.spend_capture_rate import (
     CaptureRateDay,
     CaptureRateReport,
@@ -70,7 +71,7 @@ ProviderBillingFailure: TypeAlias = ProviderBillingCredentialMissing | ProviderB
 CheckResult: TypeAlias = CaptureRateReport | ProviderBillingFailure
 
 
-class _CapturedSpendRow(BaseModel):
+class _CapturedSpendRow(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     date: str
diff --git a/litellm/proxy/spend_tracking/spend_event.py b/litellm/proxy/spend_tracking/spend_event.py
index 53f26346f85..031e84896b2 100644
--- a/litellm/proxy/spend_tracking/spend_event.py
+++ b/litellm/proxy/spend_tracking/spend_event.py
@@ -26,6 +26,7 @@ import litellm
 from litellm.proxy._types import UserAPIKeyAuth
 from litellm.proxy.db.spend_log_tool_index import response_tool_call_names
 from litellm.types.interactions import InteractionsAPIResponse
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import LiteLLMBatch, Usage
 
 SPEND_EVENT_VERSION: Final = 1
@@ -159,7 +160,7 @@ _OBJECT_MAPPING: Final = TypeAdapter(ObjectMapping)
 _COMPACT_RESPONSE: Final = TypeAdapter(CompactResponse)
 
 
-class SpendEvent(BaseModel):
+class SpendEvent(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="forbid")
 
     version: Literal[1]
diff --git a/litellm/proxy/tracing_endpoints.py b/litellm/proxy/tracing_endpoints.py
index 019829c9366..f6ddc5ddebe 100644
--- a/litellm/proxy/tracing_endpoints.py
+++ b/litellm/proxy/tracing_endpoints.py
@@ -16,7 +16,7 @@ from types import MappingProxyType
 from typing import Annotated, Final, Literal
 
 from fastapi import APIRouter, Depends, HTTPException, Query, Request, Response
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 from typing_extensions import assert_never
 
 from litellm._logging import verbose_proxy_logger
@@ -50,6 +50,7 @@ from litellm.rust_bridge.trace.generated.types import (
 from litellm.rust_bridge.trace.storage import ClickHouseStorage, Tenant
 from litellm.tracing import TraceReceiver, TracingPayloadTooLargeError
 from litellm.tracing.otlp_http import InvalidOTLPPayloadError, encode_otlp_response
+from litellm.types.llms.base import LiteLLMBaseModel
 
 router = APIRouter(tags=["agent tracing"])
 
@@ -150,7 +151,7 @@ async def ingest_otlp_traces(
     return Response(content=body, media_type=media_type)
 
 
-class TraceReadFailure(BaseModel):
+class TraceReadFailure(LiteLLMBaseModel):
     """The body of every failed trace read. Clients branch on `code`, never on `message`."""
 
     model_config = ConfigDict(frozen=True)
diff --git a/litellm/proxy/ui_crud_endpoints/latest_release_endpoints.py b/litellm/proxy/ui_crud_endpoints/latest_release_endpoints.py
index 2e90d4c0d90..720b4dd8d32 100644
--- a/litellm/proxy/ui_crud_endpoints/latest_release_endpoints.py
+++ b/litellm/proxy/ui_crud_endpoints/latest_release_endpoints.py
@@ -8,11 +8,12 @@ from typing import Annotated, Final, Literal, Protocol, TypeAlias
 
 import httpx
 from fastapi import APIRouter, Depends
-from pydantic import BaseModel, ValidationError
+from pydantic import ValidationError
 
 from litellm._logging import verbose_proxy_logger
 from litellm.caching.in_memory_cache import InMemoryCache
 from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
+from litellm.types.llms.base import LiteLLMBaseModel
 
 router: Final = APIRouter()
 
@@ -29,7 +30,7 @@ _Bucket: TypeAlias = Literal["new_features", "bug_fixes", "other_updates"]
 _PREFIX_BUCKETS: Final[Mapping[str, _Bucket]] = MappingProxyType({"feat": "new_features", "fix": "bug_fixes"})
 
 
-class LatestReleaseInfo(BaseModel):
+class LatestReleaseInfo(LiteLLMBaseModel):
     version: str
     new_features: int
     bug_fixes: int
@@ -42,7 +43,7 @@ class LatestReleaseUnavailable:
     reason: str
 
 
-class _GitHubRelease(BaseModel):
+class _GitHubRelease(LiteLLMBaseModel):
     tag_name: str
     html_url: str
     body: str
diff --git a/litellm/proxy/ui_crud_endpoints/proxy_setting_endpoints.py b/litellm/proxy/ui_crud_endpoints/proxy_setting_endpoints.py
index dc519bc4720..d307805427a 100644
--- a/litellm/proxy/ui_crud_endpoints/proxy_setting_endpoints.py
+++ b/litellm/proxy/ui_crud_endpoints/proxy_setting_endpoints.py
@@ -52,6 +52,7 @@ from litellm.repositories.table_repositories import (
 )
 from litellm.repositories.team_repository import TeamRepository
 from litellm.secret_managers.main import get_secret
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.mcp import MCPToolSearchSettings
 from litellm.types.proxy.management_endpoints.ui_sso import (
     DefaultTeamSSOParams,
@@ -168,11 +169,11 @@ def _resolve_ui_theme_field(stored_values: Mapping[str, object], field_name: str
     return env_value if _is_public_http_url(env_value) else None
 
 
-class IPAddress(BaseModel):
+class IPAddress(LiteLLMBaseModel):
     ip: str
 
 
-class UIThemeConfig(BaseModel):
+class UIThemeConfig(LiteLLMBaseModel):
     """Configuration for UI theme customization"""
 
     # Logo configuration
@@ -196,7 +197,7 @@ class UIThemeConfig(BaseModel):
     )
 
 
-class SettingsResponse(BaseModel):
+class SettingsResponse(LiteLLMBaseModel):
     """Base response model for settings with values and schema information"""
 
     values: dict[str, object]
@@ -206,7 +207,7 @@ class SettingsResponse(BaseModel):
     """Schema information including descriptions and property types for UI display"""
 
 
-class _SettingsWithSchema(BaseModel):
+class _SettingsWithSchema(LiteLLMBaseModel):
     values: dict[str, object]
     field_schema: dict[str, object]
 
@@ -233,7 +234,7 @@ class UIThemeSettingsResponse(SettingsResponse):
 _TEAM_ADMIN_FIELD_ENUM: Final = tuple(sorted(SUPPORTED_TEAM_ADMIN_PERMISSIONS))
 
 
-class UISettings(BaseModel):
+class UISettings(LiteLLMBaseModel):
     """Configuration for UI-specific flags"""
 
     model_config = ConfigDict(extra="allow")
@@ -476,7 +477,7 @@ def _get_effective_ui_settings_class() -> type[UISettings]:
     return _EFFECTIVE_UI_SETTINGS_CLASS
 
 
-class MCPSemanticFilterSettings(BaseModel):
+class MCPSemanticFilterSettings(LiteLLMBaseModel):
     """Configuration for MCP Semantic Tool Filter"""
 
     enabled: bool = Field(
@@ -512,7 +513,7 @@ class MCPToolSearchSettingsResponse(SettingsResponse):
     """Response model for native MCP tool search settings"""
 
 
-class WebSearchInterceptionSettings(BaseModel):
+class WebSearchInterceptionSettings(LiteLLMBaseModel):
     """Configuration for server-side web search interception"""
 
     enabled: bool = Field(
diff --git a/litellm/proxy/ui_crud_endpoints/user_banner_endpoints.py b/litellm/proxy/ui_crud_endpoints/user_banner_endpoints.py
index 0ff61592d9f..56339202d46 100644
--- a/litellm/proxy/ui_crud_endpoints/user_banner_endpoints.py
+++ b/litellm/proxy/ui_crud_endpoints/user_banner_endpoints.py
@@ -3,12 +3,13 @@ import json
 from typing import Annotated, Final, Literal
 
 from fastapi import APIRouter, Depends, HTTPException
-from pydantic import BaseModel, Field, ValidationError, model_validator
+from pydantic import Field, ValidationError, model_validator
 
 from litellm._uuid import uuid4
 from litellm.proxy._types import LitellmTableNames, LitellmUserRoles, UserAPIKeyAuth
 from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
 from litellm.repositories.user_banner_repository import USER_BANNER_ROW_ID, UserBannerRepository
+from litellm.types.llms.base import LiteLLMBaseModel
 
 router: Final = APIRouter()
 
@@ -17,7 +18,7 @@ USER_BANNER_MAX_MESSAGE_LENGTH: Final = 4000
 UserBannerSeverity = Literal["info", "warning", "error"]
 
 
-class UserBannerUpdate(BaseModel):
+class UserBannerUpdate(LiteLLMBaseModel):
     enabled: bool = Field(
         default=False,
         description="If true, the banner is shown to all authenticated dashboard users.",
@@ -49,7 +50,7 @@ class UserBanner(UserBannerUpdate):
     )
 
 
-class UpdateUserBannerResponse(BaseModel):
+class UpdateUserBannerResponse(LiteLLMBaseModel):
     message: str
     banner: UserBanner
 
diff --git a/litellm/rag/ingestion/gemini_ingestion.py b/litellm/rag/ingestion/gemini_ingestion.py
index a015f3cf614..4ee16685e03 100644
--- a/litellm/rag/ingestion/gemini_ingestion.py
+++ b/litellm/rag/ingestion/gemini_ingestion.py
@@ -9,7 +9,7 @@ from __future__ import annotations
 
 from typing import TYPE_CHECKING, Final, cast
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 
 from litellm._logging import verbose_logger
 from litellm.llms.custom_httpx.http_handler import (
@@ -19,13 +19,14 @@ from litellm.llms.custom_httpx.http_handler import (
 )
 from litellm.llms.gemini.common_utils import GeminiModelInfo
 from litellm.rag.ingestion.base_ingestion import BaseRAGIngestion
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     from litellm import Router
     from litellm.types.rag import RAGIngestOptions
 
 
-class _WhiteSpaceConfig(BaseModel):
+class _WhiteSpaceConfig(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore", frozen=True, hide_input_in_errors=True)
 
     max_tokens_per_chunk: object = 800
diff --git a/litellm/responses/additional_tools.py b/litellm/responses/additional_tools.py
index 5239bd395cc..4b897ff58c0 100644
--- a/litellm/responses/additional_tools.py
+++ b/litellm/responses/additional_tools.py
@@ -2,19 +2,20 @@ from collections.abc import Sequence
 from dataclasses import dataclass
 from typing import Final, cast  # noqa: TID251  # validating the openai tool union strips vendor keys from raw tools
 
-from pydantic import BaseModel, ValidationError
+from pydantic import ValidationError
 
 from litellm._logging import verbose_logger
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import ALL_RESPONSES_API_TOOL_PARAMS, ResponseInputParam
 
 ADDITIONAL_TOOLS_INPUT_ITEM_TYPE: Final = "additional_tools"
 
 
-class _InputItemType(BaseModel):
+class _InputItemType(LiteLLMBaseModel):
     type: str = ""
 
 
-class _AdditionalToolsItem(BaseModel):
+class _AdditionalToolsItem(LiteLLMBaseModel):
     tools: tuple[dict[str, object], ...] = ()
 
 
diff --git a/litellm/responses/litellm_completion_transformation/custom_tools.py b/litellm/responses/litellm_completion_transformation/custom_tools.py
index 7888a07e248..1943dda09b3 100644
--- a/litellm/responses/litellm_completion_transformation/custom_tools.py
+++ b/litellm/responses/litellm_completion_transformation/custom_tools.py
@@ -20,8 +20,9 @@ from collections.abc import Mapping, Sequence
 from types import MappingProxyType
 from typing import Final
 
-from pydantic import BaseModel, TypeAdapter, ValidationError
+from pydantic import TypeAdapter, ValidationError
 
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import (
     ChatCompletionToolParam,
     ChatCompletionToolParamFunctionChunk,
@@ -39,7 +40,7 @@ def openai_shaped_tool_call_item_id(item_type: str, tool_id: str) -> str:
     return f"{prefix}_{tool_id}"
 
 
-class _ToolNameFields(BaseModel):
+class _ToolNameFields(LiteLLMBaseModel):
     type: str = ""
     name: str = ""
     tools: tuple[object, ...] = ()
@@ -151,7 +152,7 @@ def build_tool_call_item_kwargs(
     return kwargs
 
 
-class _CustomToolFormat(BaseModel):
+class _CustomToolFormat(LiteLLMBaseModel):
     syntax: str = ""
     definition: str = ""
 
diff --git a/litellm/router_strategy/auto_router/auto_router.py b/litellm/router_strategy/auto_router/auto_router.py
index 250201a46d1..6b97276a079 100644
--- a/litellm/router_strategy/auto_router/auto_router.py
+++ b/litellm/router_strategy/auto_router/auto_router.py
@@ -6,7 +6,7 @@ import asyncio
 from collections.abc import Mapping, Sequence
 from typing import TYPE_CHECKING, Any, Final, Optional
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 
 from litellm._logging import verbose_router_logger
 from litellm.constants import DEFAULT_AUTO_ROUTER_MAX_INPUT_CHARS
@@ -16,6 +16,7 @@ from litellm.litellm_core_utils.internal_call_metadata import (
     forwarded_internal_call_metadata,
     parent_session_kwargs,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import AUTOROUTER_CLASSIFIER_CALL_ORIGIN
 
 if TYPE_CHECKING:
@@ -33,7 +34,7 @@ else:
     LiteLLMRouterEncoder = Any
 
 
-class _CallerMetadata(BaseModel):
+class _CallerMetadata(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="ignore")
 
     metadata: Mapping[str, object] | None = None
diff --git a/litellm/router_strategy/complexity_router/capability_classifier.py b/litellm/router_strategy/complexity_router/capability_classifier.py
index 93077af9e47..717e8d389a3 100644
--- a/litellm/router_strategy/complexity_router/capability_classifier.py
+++ b/litellm/router_strategy/complexity_router/capability_classifier.py
@@ -9,7 +9,9 @@ from sys import float_info
 from types import MappingProxyType
 from typing import Final, Literal, NamedTuple, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, Field, StrictFloat, TypeAdapter, model_validator
+from pydantic import ConfigDict, Field, StrictFloat, TypeAdapter, model_validator
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 CapabilityBoundary: TypeAlias = Literal["supported", "uncertain", "unsupported", "unmatched"]
 CapabilityRule: TypeAlias = Literal[
@@ -111,7 +113,7 @@ _RULE_BOUNDARIES: Final = MappingProxyType(
 )
 
 
-class CapabilityClassifierVerdict(BaseModel):
+class CapabilityClassifierVerdict(LiteLLMBaseModel):
     """Strict structured verdict returned by the capability forecaster."""
 
     model_config = ConfigDict(extra="forbid", frozen=True)
diff --git a/litellm/router_strategy/complexity_router/complexity_router.py b/litellm/router_strategy/complexity_router/complexity_router.py
index 43200271f9f..5600f4c9354 100644
--- a/litellm/router_strategy/complexity_router/complexity_router.py
+++ b/litellm/router_strategy/complexity_router/complexity_router.py
@@ -65,6 +65,7 @@ from litellm.router_strategy.complexity_router.tier_predictor import (
 )
 from litellm.router_utils.pre_call_checks.deployment_affinity_check import DeploymentAffinityCheck
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.custom_http import httpxSpecialProvider
 from litellm.types.llms.openai import (
     AllMessageValues,
@@ -136,13 +137,13 @@ else:
     SemanticRouter = Any
 
 
-class TierClassification(BaseModel):
+class TierClassification(LiteLLMBaseModel):
     """Structured response schema for the LLM-based complexity classifier."""
 
     tier: Literal["SIMPLE", "MEDIUM", "COMPLEX", "REASONING"]
 
 
-class _LabeledTierClassification(BaseModel):
+class _LabeledTierClassification(LiteLLMBaseModel):
     """Parses the classifier's reply when the wire carries operator-chosen tier strings."""
 
     tier: str
diff --git a/litellm/router_strategy/complexity_router/config.py b/litellm/router_strategy/complexity_router/config.py
index 8d00b45fed0..bd067892ddc 100644
--- a/litellm/router_strategy/complexity_router/config.py
+++ b/litellm/router_strategy/complexity_router/config.py
@@ -15,7 +15,6 @@ from types import MappingProxyType
 from typing import Annotated, Final, Literal, NamedTuple
 
 from pydantic import (
-    BaseModel,
     ConfigDict,
     Field,
     SkipValidation,
@@ -26,6 +25,8 @@ from pydantic import (
     model_validator,
 )
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 with warnings.catch_warnings():
     warnings.simplefilter("ignore", DeprecationWarning)
     import sre_constants
@@ -96,7 +97,7 @@ DEFAULT_CLASSIFIER_CONTEXT_WINDOW_SIZE: Final[int] = 3
 DEFAULT_CLASSIFIER_CONTEXT_BUDGET_CHARS: Final[int] = 8000
 
 
-class KeywordTierRule(BaseModel):
+class KeywordTierRule(LiteLLMBaseModel):
     """A deterministic override: if any keyword matches, route to this tier."""
 
     keywords: list[str] = Field(
@@ -173,7 +174,7 @@ def normalize_classification_examples(value: str | None) -> str | None:
 _BUILT_IN_TIER_NAMES: Final[str] = ", ".join(ComplexityTier.__members__)
 
 
-class TierDefinition(BaseModel):
+class TierDefinition(LiteLLMBaseModel):
     """An operator-defined tier: the name the LLM classifier must return and its rubric description."""
 
     name: str = Field(
@@ -217,7 +218,7 @@ class TierDefinition(BaseModel):
         return self
 
 
-class ReminderMarkerPair(BaseModel):
+class ReminderMarkerPair(LiteLLMBaseModel):
     """One open/close delimiter pair a harness wraps injected context in.
 
     Normalizing here rather than at the scan is what makes matching case-insensitive: markers reach
@@ -241,7 +242,7 @@ class ReminderMarkerPair(BaseModel):
         return self
 
 
-class ComplexityTierModel(BaseModel):
+class ComplexityTierModel(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     model_name: str
@@ -477,7 +478,7 @@ DEFAULT_TIER_MODELS: Final[dict[str, str]] = {
 }
 
 
-class ClassifierVisionConfig(BaseModel):
+class ClassifierVisionConfig(LiteLLMBaseModel):
     """Whether the LLM classifier sees the images on the request it is classifying.
 
     Off by default because images cost far more than the text ask they arrive with, and the
@@ -508,7 +509,7 @@ class ClassifierVisionConfig(BaseModel):
     )
 
 
-class ClassifierLLMConfig(BaseModel):
+class ClassifierLLMConfig(LiteLLMBaseModel):
     """Configuration for the LLM-based complexity classifier."""
 
     model: str = Field(
@@ -604,7 +605,7 @@ class ClassifierLLMConfig(BaseModel):
         return self
 
 
-class CapabilityCalibrationConfig(BaseModel):
+class CapabilityCalibrationConfig(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     version: str = Field(min_length=1, max_length=128, pattern=r"^\S(?:.*\S)?$")
@@ -617,7 +618,7 @@ class CapabilityCalibrationConfig(BaseModel):
         return 1.0 / (1.0 + math.exp(-log_odds))
 
 
-class CapabilityClassifierConfig(BaseModel):
+class CapabilityClassifierConfig(LiteLLMBaseModel):
     """Switchyard-compatible probability threshold policy for two model tiers."""
 
     model_config = ConfigDict(extra="forbid", frozen=True)
@@ -695,7 +696,7 @@ def normalize_classifier_config_aliases(config: Mapping[str, object]) -> Mapping
     return normalized
 
 
-class OpenSourceClassifierConfig(BaseModel):
+class OpenSourceClassifierConfig(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     provider: Literal["jev", "laya", "bespoke"] = "jev"
@@ -904,7 +905,7 @@ def custom_pattern_work(pattern: str) -> int | str:
     return cost if isinstance(cost, str) else cost.steps
 
 
-class CustomDimension(BaseModel):
+class CustomDimension(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     name: str = Field(min_length=1, max_length=64, pattern=r"^[A-Za-z][A-Za-z0-9_]*$")
@@ -940,7 +941,7 @@ class CustomDimension(BaseModel):
         )
 
 
-class ContextCompactionConfig(BaseModel):
+class ContextCompactionConfig(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="forbid")
 
     model: str | None = Field(default=None, min_length=1)
@@ -949,7 +950,7 @@ class ContextCompactionConfig(BaseModel):
     timeout_seconds: float = Field(default=120, gt=0)
 
 
-class ComplexityRouterConfig(BaseModel):
+class ComplexityRouterConfig(LiteLLMBaseModel):
     """Configuration for the ComplexityRouter."""
 
     @model_validator(mode="before")
diff --git a/litellm/router_strategy/complexity_router/fuse_presets.py b/litellm/router_strategy/complexity_router/fuse_presets.py
index 66a96ec5ad5..3bb0f2b41d8 100644
--- a/litellm/router_strategy/complexity_router/fuse_presets.py
+++ b/litellm/router_strategy/complexity_router/fuse_presets.py
@@ -2,12 +2,14 @@ from functools import lru_cache
 from importlib.resources import files
 from typing import Annotated, Final, Literal, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, Field, StringConstraints
+from pydantic import ConfigDict, Field, StringConstraints
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 ProfileText: TypeAlias = Annotated[str, StringConstraints(strip_whitespace=True, min_length=1, max_length=4000)]
 
 
-class FuseModelPreset(BaseModel):
+class FuseModelPreset(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     id: str
@@ -17,7 +19,7 @@ class FuseModelPreset(BaseModel):
     model: str
 
 
-class FuseHarnessPreset(BaseModel):
+class FuseHarnessPreset(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     id: str
@@ -26,7 +28,7 @@ class FuseHarnessPreset(BaseModel):
     sources: tuple[str, ...] = Field(min_length=1)
 
 
-class FusePresetCatalog(BaseModel):
+class FusePresetCatalog(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     version: str
diff --git a/litellm/router_strategy/complexity_router/jev_classifier.py b/litellm/router_strategy/complexity_router/jev_classifier.py
index 2b97ae824dd..93c8b0c1bd3 100644
--- a/litellm/router_strategy/complexity_router/jev_classifier.py
+++ b/litellm/router_strategy/complexity_router/jev_classifier.py
@@ -5,7 +5,7 @@ from typing import Annotated, Final, Literal, NamedTuple, Protocol, TypeAlias
 from uuid import uuid4
 
 import httpx
-from pydantic import BaseModel, ConfigDict, Field, TypeAdapter, ValidationError
+from pydantic import ConfigDict, Field, TypeAdapter, ValidationError
 
 import litellm
 from litellm._logging import verbose_router_logger
@@ -23,13 +23,14 @@ from litellm.proxy.pass_through_endpoints.llm_provider_handlers.typesafe_passthr
     TypeSafePassthroughLoggingHandler,
 )
 from litellm.router_strategy.complexity_router.config import DEFAULT_JEV_INSTRUCTIONS as _DEFAULT_JEV_INSTRUCTIONS
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import AUTOROUTER_CLASSIFIER_CALL_ORIGIN
 
 JevProbability: TypeAlias = Annotated[float, Field(ge=0.0, le=1.0)]
 DEFAULT_JEV_INSTRUCTIONS: Final = _DEFAULT_JEV_INSTRUCTIONS
 
 
-class JevChoiceQuestion(BaseModel):
+class JevChoiceQuestion(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     type: Literal["choice"] = "choice"
@@ -37,7 +38,7 @@ class JevChoiceQuestion(BaseModel):
     criteria: Mapping[str, str]
 
 
-class JevSystemOneRequest(BaseModel):
+class JevSystemOneRequest(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     state: str
@@ -45,7 +46,7 @@ class JevSystemOneRequest(BaseModel):
     questions: Mapping[str, JevChoiceQuestion]
 
 
-class JevChoiceAnswer(BaseModel):
+class JevChoiceAnswer(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, allow_inf_nan=False)
 
     type: Literal["choice"]
@@ -54,14 +55,14 @@ class JevChoiceAnswer(BaseModel):
     confidence: JevProbability
 
 
-class JevUsage(BaseModel):
+class JevUsage(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     input_tokens: int = Field(default=0, ge=0, strict=True)
     output_tokens: int = Field(default=0, ge=0, strict=True)
 
 
-class JevSystemOneResponse(BaseModel):
+class JevSystemOneResponse(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     model: str | None = None
@@ -204,7 +205,7 @@ class JevVerdict(NamedTuple):
     provider: Literal["typesafe", "laya", "bespoke"] = "typesafe"
 
 
-class _RegistryPricing(BaseModel):
+class _RegistryPricing(LiteLLMBaseModel):
     input_cost_per_token: float = 0.0
     output_cost_per_token: float = 0.0
 
diff --git a/litellm/router_strategy/complexity_router/llm_v2.py b/litellm/router_strategy/complexity_router/llm_v2.py
index 8ef2f554ab2..c63d05de72b 100644
--- a/litellm/router_strategy/complexity_router/llm_v2.py
+++ b/litellm/router_strategy/complexity_router/llm_v2.py
@@ -7,13 +7,14 @@ from dataclasses import dataclass
 from sys import float_info
 from typing import Annotated, Final, Literal, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, Field, StrictFloat, StringConstraints, TypeAdapter, model_validator
+from pydantic import ConfigDict, Field, StrictFloat, StringConstraints, TypeAdapter, model_validator
 from typing_extensions import ReadOnly, TypedDict
 
 from litellm.llms.base_llm.base_utils import (
     type_to_response_format_param,  # pyright: ignore[reportUnknownVariableType]  # legacy output validated below
 )
 from litellm.router_strategy.complexity_router.fuse_presets import ProfileText, resolve_fuse_profile
+from litellm.types.llms.base import LiteLLMBaseModel
 
 ShortText: TypeAlias = Annotated[str, StringConstraints(strip_whitespace=True, min_length=1, max_length=512)]
 VerdictText: TypeAlias = Annotated[str, StringConstraints(strip_whitespace=True, min_length=1)]
@@ -80,7 +81,7 @@ these forecasts are calibrated. Do not optimize cost or output a selected model.
 Return only JSON matching the response schema. Keep text fields concise."""
 
 
-class LLMV2Demands(BaseModel):
+class LLMV2Demands(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     reasoning: Literal["routine", "multistep", "open_ended", "unknown"]
@@ -88,21 +89,21 @@ class LLMV2Demands(BaseModel):
     specification: Literal["clear", "ambiguous", "unknown"]
 
 
-class LLMV2SolverForecast(BaseModel):
+class LLMV2SolverForecast(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     likely_failure: VerdictText
     p_solve: StrictFloat = Field(ge=0.0, le=1.0)
 
 
-class LLMV2SolverForecasts(BaseModel):
+class LLMV2SolverForecasts(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     efficient: LLMV2SolverForecast
     capable: LLMV2SolverForecast
 
 
-class LLMV2Verdict(BaseModel):
+class LLMV2Verdict(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     crux: VerdictText
@@ -111,7 +112,7 @@ class LLMV2Verdict(BaseModel):
     forecasts: LLMV2SolverForecasts
 
 
-class LLMV2ProbabilityCalibration(BaseModel):
+class LLMV2ProbabilityCalibration(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     slope: float = Field(gt=0.0, allow_inf_nan=False)
@@ -126,7 +127,7 @@ class LLMV2ProbabilityCalibration(BaseModel):
         return exponential / (1.0 + exponential)
 
 
-class LLMV2Calibration(BaseModel):
+class LLMV2Calibration(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     version: ShortText
@@ -135,7 +136,7 @@ class LLMV2Calibration(BaseModel):
     capable: LLMV2ProbabilityCalibration
 
 
-class LLMV2Config(BaseModel):
+class LLMV2Config(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid", frozen=True)
 
     efficient_tier: str = "SIMPLE"
diff --git a/litellm/router_strategy/complexity_router/tier_predictor.py b/litellm/router_strategy/complexity_router/tier_predictor.py
index 7775c36e795..246e58dbadf 100644
--- a/litellm/router_strategy/complexity_router/tier_predictor.py
+++ b/litellm/router_strategy/complexity_router/tier_predictor.py
@@ -7,12 +7,13 @@ from pathlib import Path
 from types import MappingProxyType
 from typing import Final, Literal
 
-from pydantic import BaseModel, Field, model_validator
+from pydantic import Field, model_validator
 
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.router import RequestType
 
 
-class TierGlobalStatistic(BaseModel):
+class TierGlobalStatistic(LiteLLMBaseModel):
     tier: int = Field(ge=1, le=4)
     successes: float = Field(ge=0.0)
     observations: float = Field(gt=0.0)
@@ -32,7 +33,7 @@ class TierCohortStatistic(TierGlobalStatistic):
     cohort: str = Field(min_length=1)
 
 
-class TierDataset(BaseModel):
+class TierDataset(LiteLLMBaseModel):
     name: str = Field(min_length=1)
     url: str = Field(min_length=1)
     license: str = Field(min_length=1)
@@ -40,7 +41,7 @@ class TierDataset(BaseModel):
     success_definition: str = Field(default="quality score meets the dataset success threshold", min_length=1)
 
 
-class TrainedTierArtifact(BaseModel):
+class TrainedTierArtifact(LiteLLMBaseModel):
     schema_version: Literal[1] = 1
     global_statistics: tuple[TierGlobalStatistic, ...]
     domain_statistics: tuple[TierDomainStatistic, ...] = ()
diff --git a/litellm/router_strategy/quality_router/config.py b/litellm/router_strategy/quality_router/config.py
index e07a6962d72..6e04a804c34 100644
--- a/litellm/router_strategy/quality_router/config.py
+++ b/litellm/router_strategy/quality_router/config.py
@@ -4,7 +4,9 @@ Configuration models for the QualityRouter.
 
 from typing import Final
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 # Default mapping from ComplexityTier name (string) to quality tier (int).
 # Higher tier = higher capability requirement.
@@ -16,7 +18,7 @@ DEFAULT_COMPLEXITY_TO_QUALITY: Final[dict[str, int]] = {
 }
 
 
-class QualityRouterConfig(BaseModel):
+class QualityRouterConfig(LiteLLMBaseModel):
     """Configuration for the QualityRouter."""
 
     available_models: list[str] = Field(
@@ -40,7 +42,7 @@ class QualityRouterConfig(BaseModel):
     model_config = ConfigDict(extra="allow")
 
 
-class RoutingPreferences(BaseModel):
+class RoutingPreferences(LiteLLMBaseModel):
     """Per-deployment routing preferences declared on model_info."""
 
     quality_tier: int = Field(
diff --git a/litellm/rust_bridge/trace/generated/models.py b/litellm/rust_bridge/trace/generated/models.py
index db7b76eba96..ea2c8bda648 100644
--- a/litellm/rust_bridge/trace/generated/models.py
+++ b/litellm/rust_bridge/trace/generated/models.py
@@ -4,10 +4,12 @@ from __future__ import annotations
 
 from typing import Annotated, Literal, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class ActivityAvailability(BaseModel):
+class ActivityAvailability(LiteLLMBaseModel):
     model_config = ConfigDict(
         frozen=True,
     )
@@ -16,7 +18,7 @@ class ActivityAvailability(BaseModel):
     requests: bool = False
 
 
-class AgentRow(BaseModel):
+class AgentRow(LiteLLMBaseModel):
     model_config = ConfigDict(
         frozen=True,
     )
@@ -57,7 +59,7 @@ Count1: TypeAlias = Annotated[
 ]
 
 
-class CountRow(BaseModel):
+class CountRow(LiteLLMBaseModel):
     model_config = ConfigDict(
         frozen=True,
     )
@@ -170,7 +172,7 @@ Selected1: TypeAlias = Annotated[
 ]
 
 
-class ExecutionRow(BaseModel):
+class ExecutionRow(LiteLLMBaseModel):
     model_config = ConfigDict(
         frozen=True,
     )
@@ -190,7 +192,7 @@ class ExecutionRow(BaseModel):
     selection_key: str = ""
 
 
-class LensAccessParams(BaseModel):
+class LensAccessParams(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
@@ -201,7 +203,7 @@ class LensAccessParams(BaseModel):
     key_hash: str
 
 
-class LensContentParams(BaseModel):
+class LensContentParams(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
@@ -218,7 +220,7 @@ class LensContentParams(BaseModel):
     offset: int = Field(..., ge=0, le=4294967295)
 
 
-class LensEvidenceParams(BaseModel):
+class LensEvidenceParams(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
@@ -238,7 +240,7 @@ class LensEvidenceParams(BaseModel):
 ExecutionSource: TypeAlias = Literal["traces", "requests", "both"]
 
 
-class LensSampleParams(BaseModel):
+class LensSampleParams(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
@@ -264,7 +266,7 @@ class LensSampleParams(BaseModel):
     offset: int = Field(..., ge=0, le=18446744073709551615)
 
 
-class PartRow(BaseModel):
+class PartRow(LiteLLMBaseModel):
     model_config = ConfigDict(
         frozen=True,
     )
@@ -282,7 +284,7 @@ class PartRow(BaseModel):
 TraceTableName: TypeAlias = Literal["otel_traces", "agent_traces_by_key", "spend_logs"]
 
 
-class TraceQueryColumn(BaseModel):
+class TraceQueryColumn(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="allow",
         frozen=True,
@@ -292,7 +294,7 @@ class TraceQueryColumn(BaseModel):
     type: str
 
 
-class TraceQueryNormalizedField(BaseModel):
+class TraceQueryNormalizedField(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
@@ -317,7 +319,7 @@ MetadataValueType: TypeAlias = Literal["array", "boolean", "integer", "null", "n
 MapValueType: TypeAlias = Literal["String"]
 
 
-class TraceQueryRelationship(BaseModel):
+class TraceQueryRelationship(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
@@ -329,7 +331,7 @@ class TraceQueryRelationship(BaseModel):
     meaning: str
 
 
-class TraceQueryExample(BaseModel):
+class TraceQueryExample(LiteLLMBaseModel):
     model_config = ConfigDict(
         frozen=True,
     )
@@ -338,7 +340,7 @@ class TraceQueryExample(BaseModel):
     sql: str
 
 
-class TraceQueryTable(BaseModel):
+class TraceQueryTable(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
@@ -348,7 +350,7 @@ class TraceQueryTable(BaseModel):
     columns: tuple[TraceQueryColumn, ...]
 
 
-class TraceQueryMetadataField(BaseModel):
+class TraceQueryMetadataField(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
@@ -359,7 +361,7 @@ class TraceQueryMetadataField(BaseModel):
     expression: str
 
 
-class TraceQueryAttributeField(BaseModel):
+class TraceQueryAttributeField(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
@@ -370,7 +372,7 @@ class TraceQueryAttributeField(BaseModel):
     expression: str
 
 
-class TraceQueryMetadata(BaseModel):
+class TraceQueryMetadata(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
@@ -387,7 +389,7 @@ class TraceQueryMetadata(BaseModel):
     scope: str
 
 
-class TraceQueryAttributes(BaseModel):
+class TraceQueryAttributes(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
@@ -402,7 +404,7 @@ class TraceQueryAttributes(BaseModel):
     scope: str
 
 
-class TraceQueryHelp(BaseModel):
+class TraceQueryHelp(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
diff --git a/litellm/rust_bridge/trace/generated/requests.py b/litellm/rust_bridge/trace/generated/requests.py
index dd5304ea09b..9c30049ce14 100644
--- a/litellm/rust_bridge/trace/generated/requests.py
+++ b/litellm/rust_bridge/trace/generated/requests.py
@@ -4,10 +4,12 @@ from __future__ import annotations
 
 from typing import Annotated, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class TraceDetailRequest(BaseModel):
+class TraceDetailRequest(LiteLLMBaseModel):
     model_config = ConfigDict(
         frozen=True,
     )
@@ -17,7 +19,7 @@ class TraceDetailRequest(BaseModel):
     page_size: int | None = Field(None, ge=1, le=500)
 
 
-class TraceErrorPageRequest(BaseModel):
+class TraceErrorPageRequest(LiteLLMBaseModel):
     model_config = ConfigDict(
         frozen=True,
     )
@@ -26,7 +28,7 @@ class TraceErrorPageRequest(BaseModel):
     cursor: str | None = Field(None, max_length=512)
 
 
-class TraceListRequest(BaseModel):
+class TraceListRequest(LiteLLMBaseModel):
     model_config = ConfigDict(
         frozen=True,
     )
@@ -36,7 +38,7 @@ class TraceListRequest(BaseModel):
     cursor: str | None = Field(None, max_length=512)
 
 
-class TraceQueryRequest(BaseModel):
+class TraceQueryRequest(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
@@ -45,7 +47,7 @@ class TraceQueryRequest(BaseModel):
     sql: str
 
 
-class TraceSpanRequest(BaseModel):
+class TraceSpanRequest(LiteLLMBaseModel):
     model_config = ConfigDict(
         frozen=True,
     )
diff --git a/litellm/rust_bridge/trace/generated/responses.py b/litellm/rust_bridge/trace/generated/responses.py
index b6de931c7ad..38dee46b305 100644
--- a/litellm/rust_bridge/trace/generated/responses.py
+++ b/litellm/rust_bridge/trace/generated/responses.py
@@ -5,10 +5,12 @@ from __future__ import annotations
 from collections.abc import Mapping
 from typing import Annotated, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, Field, JsonValue
+from pydantic import ConfigDict, Field, JsonValue
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class TraceSQLResponse(BaseModel):
+class TraceSQLResponse(LiteLLMBaseModel):
     model_config = ConfigDict(
         extra="forbid",
         frozen=True,
diff --git a/litellm/rust_bridge/trace/queries.py b/litellm/rust_bridge/trace/queries.py
index 681b6af9f7e..42d6430782f 100644
--- a/litellm/rust_bridge/trace/queries.py
+++ b/litellm/rust_bridge/trace/queries.py
@@ -4,6 +4,8 @@ from typing import Final, Generic, TypeVar
 
 from pydantic import BaseModel, ConfigDict, JsonValue, TypeAdapter
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 from .generated.models import (
     ActivityAvailability,
     AgentRow,
@@ -21,14 +23,14 @@ from .generated.types import ReadQueryName
 _RESPONSE_CONFIG: Final = ConfigDict(frozen=True, extra="allow")
 
 
-class TraceQueryStatistics(BaseModel):
+class TraceQueryStatistics(LiteLLMBaseModel):
     model_config = _RESPONSE_CONFIG
     elapsed: float
     rows_read: int | str
     bytes_read: int | str
 
 
-class ClickHouseSQLEnvelope(BaseModel):
+class ClickHouseSQLEnvelope(LiteLLMBaseModel):
     model_config = _RESPONSE_CONFIG
     meta: tuple[TraceQueryColumn, ...]
     data: tuple[Mapping[str, JsonValue], ...]
@@ -40,7 +42,7 @@ ParamsT: Final = TypeVar("ParamsT", bound=BaseModel)
 RowT: Final = TypeVar("RowT")
 
 
-class QueryResponse(BaseModel, Generic[RowT]):
+class QueryResponse(LiteLLMBaseModel, Generic[RowT]):
     model_config = ConfigDict(frozen=True)
     data: tuple[RowT, ...]
 
diff --git a/litellm/scheduler.py b/litellm/scheduler.py
index e19e386f527..c88cc4ce0c0 100644
--- a/litellm/scheduler.py
+++ b/litellm/scheduler.py
@@ -2,12 +2,11 @@ import enum
 import heapq
 from typing import Final
 
-from pydantic import BaseModel
-
 from litellm import print_verbose
 from litellm._internal_context import with_service_target
 from litellm.caching.caching import DualCache, RedisCache
 from litellm.constants import DEFAULT_IN_MEMORY_TTL, DEFAULT_POLLING_INTERVAL
+from litellm.types.llms.base import LiteLLMBaseModel
 
 SCHEDULER_QUEUE_TARGET: Final = "scheduler_queue"
 
@@ -17,7 +16,7 @@ class SchedulerCacheKeys(enum.Enum):
     default_in_memory_ttl = DEFAULT_IN_MEMORY_TTL  # cache queue in-memory for 5s when redis cache available
 
 
-class FlowItem(BaseModel):
+class FlowItem(LiteLLMBaseModel):
     priority: int  # Priority between 0 and 255
     request_id: str
     model_name: str
diff --git a/litellm/secret_managers/main.py b/litellm/secret_managers/main.py
index 275c029b0b7..a8420d3cbc6 100644
--- a/litellm/secret_managers/main.py
+++ b/litellm/secret_managers/main.py
@@ -6,7 +6,7 @@ import traceback
 from typing import Final
 
 import httpx
-from pydantic import BaseModel, TypeAdapter, ValidationError
+from pydantic import TypeAdapter, ValidationError
 
 import litellm
 from litellm._logging import verbose_logger
@@ -16,6 +16,7 @@ from litellm.secret_managers.dispatch import get_secret_from_manager
 from litellm.secret_managers.get_azure_ad_token_provider import (
     get_azure_ad_token_provider,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 
 oidc_cache: Final = DualCache()
 
@@ -25,7 +26,7 @@ _PARSED_LITERAL: Final = TypeAdapter(object)
 _OIDC_TOKEN_EXPIRY_MARGIN_SECONDS: Final = 60
 
 
-class _OidcTokenClaims(BaseModel):
+class _OidcTokenClaims(LiteLLMBaseModel):
     exp: float | None = None
 
 
diff --git a/litellm/types/access_group.py b/litellm/types/access_group.py
index 951e5a414b4..b4bf345ecbe 100644
--- a/litellm/types/access_group.py
+++ b/litellm/types/access_group.py
@@ -1,9 +1,9 @@
 from datetime import datetime
 
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class AccessGroupCreateRequest(BaseModel):
+class AccessGroupCreateRequest(LiteLLMBaseModel):
     access_group_name: str
     description: str | None = None
     access_model_names: list[str] | None = None
@@ -13,7 +13,7 @@ class AccessGroupCreateRequest(BaseModel):
     assigned_key_ids: list[str] | None = None
 
 
-class AccessGroupUpdateRequest(BaseModel):
+class AccessGroupUpdateRequest(LiteLLMBaseModel):
     access_group_name: str | None = None
     description: str | None = None
     access_model_names: list[str] | None = None
@@ -23,14 +23,14 @@ class AccessGroupUpdateRequest(BaseModel):
     assigned_key_ids: list[str] | None = None
 
 
-class AccessGroupResource(BaseModel):
+class AccessGroupResource(LiteLLMBaseModel):
     """A resource referenced by an access group. `name` is null when the id no longer resolves or has no alias."""
 
     id: str
     name: str | None
 
 
-class AccessGroupResponse(BaseModel):
+class AccessGroupResponse(LiteLLMBaseModel):
     access_group_id: str
     access_group_name: str
     description: str | None = None
diff --git a/litellm/types/agents.py b/litellm/types/agents.py
index 94adb9f7c4a..badaf20dba0 100644
--- a/litellm/types/agents.py
+++ b/litellm/types/agents.py
@@ -3,10 +3,10 @@ from datetime import datetime
 from typing import TYPE_CHECKING, Annotated, Any, Final, Literal, TypeAlias
 from urllib.parse import urlsplit
 
-from pydantic import BaseModel, ConfigDict, Field, PrivateAttr, StrictInt, field_validator
+from pydantic import ConfigDict, Field, PrivateAttr, StrictInt, field_validator
 from typing_extensions import ReadOnly, Required, TypedDict
 
-from litellm.types.llms.base import LiteLLMPydanticObjectBase
+from litellm.types.llms.base import LiteLLMBaseModel, LiteLLMPydanticObjectBase
 from litellm.types.proxy.agent_identity import (
     AgentExecutionMode,
     AgentIdentityBinding,
@@ -184,14 +184,14 @@ class AgentObjectPermission(TypedDict, total=False):
     agents: list[str] | None
 
 
-class AgentKillSwitchBearerAuth(BaseModel):
+class AgentKillSwitchBearerAuth(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="forbid")
 
     type: Literal["bearer"]
     token: str
 
 
-class AgentKillSwitchApiKeyAuth(BaseModel):
+class AgentKillSwitchApiKeyAuth(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="forbid")
 
     type: Literal["api_key"]
@@ -199,7 +199,7 @@ class AgentKillSwitchApiKeyAuth(BaseModel):
     api_key: str
 
 
-class AgentKillSwitchBasicAuth(BaseModel):
+class AgentKillSwitchBasicAuth(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="forbid")
 
     type: Literal["basic"]
@@ -215,7 +215,7 @@ AgentKillSwitchAuth: TypeAlias = Annotated[
 AgentKillSwitchMethod: TypeAlias = Literal["POST", "PUT", "PATCH", "DELETE", "GET"]
 
 
-class AgentKillSwitchConfig(BaseModel):
+class AgentKillSwitchConfig(LiteLLMBaseModel):
     """Webhook an admin fires to shut an agent down out of band. LiteLLM only
     makes the call; whatever the endpoint does with it is the agent's business."""
 
@@ -237,7 +237,7 @@ class AgentKillSwitchConfig(BaseModel):
         return value
 
 
-class AgentKillSwitchResult(BaseModel):
+class AgentKillSwitchResult(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     agent_id: str
@@ -292,7 +292,7 @@ AGENT_CALLER_USER_ID_HEADER: Final = "x-litellm-user-id"
 AGENT_CALLER_TEAM_ID_HEADER: Final = "x-litellm-team-id"
 
 
-class AgentCaller(BaseModel):
+class AgentCaller(LiteLLMBaseModel):
     """The user and team that invoked an agent, echoed back by the agent on its own proxy calls.
     Only ever narrows what the agent's key may do."""
 
@@ -305,13 +305,13 @@ class AgentCaller(BaseModel):
 # Request/Response models for CRUD endpoints
 
 
-class AgentKeySummary(BaseModel):
+class AgentKeySummary(LiteLLMBaseModel):
     token: str
     key_alias: str | None = None
     key_name: str | None = None
 
 
-class AgentResponse(BaseModel):
+class AgentResponse(LiteLLMBaseModel):
     identity: AgentIdentityBinding | None = None
     identity_managed: bool = False
     enabled: bool = True
@@ -339,7 +339,7 @@ class AgentResponse(BaseModel):
     updated_by: str | None = None
 
 
-class ListAgentsResponse(BaseModel):
+class ListAgentsResponse(LiteLLMBaseModel):
     agents: list[AgentResponse]
 
 
@@ -403,13 +403,13 @@ class AgentVersionsResponse(LiteLLMPydanticObjectBase):
     _hidden_params: dict[str, object] = PrivateAttr(default_factory=dict)
 
 
-class AgentMakePublicResponse(BaseModel):
+class AgentMakePublicResponse(LiteLLMBaseModel):
     message: str
     public_agent_groups: list[str]
     updated_by: str
 
 
-class MakeAgentsPublicRequest(BaseModel):
+class MakeAgentsPublicRequest(LiteLLMBaseModel):
     agent_ids: list[str]
 
 
diff --git a/litellm/types/caching.py b/litellm/types/caching.py
index 6d42556a9c4..036f7c452ea 100644
--- a/litellm/types/caching.py
+++ b/litellm/types/caching.py
@@ -2,9 +2,10 @@ from collections.abc import Sequence
 from enum import Enum
 from typing import Any, Final, Literal, Optional, Union
 
-from pydantic import BaseModel
 from typing_extensions import ReadOnly, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 
 class LiteLLMCacheType(str, Enum):
     LOCAL = "local"
@@ -114,7 +115,7 @@ DynamicCacheControl = TypedDict(
 )
 
 
-class CachePingResponse(BaseModel):
+class CachePingResponse(LiteLLMBaseModel):
     status: str
     cache_type: str
     ping_response: bool | None = None
@@ -125,7 +126,7 @@ class CachePingResponse(BaseModel):
     health_check_cache_params: dict | None = None
 
 
-class HealthCheckCacheParams(BaseModel):
+class HealthCheckCacheParams(LiteLLMBaseModel):
     """
     Cache Params returned on /cache/ping call
     """
diff --git a/litellm/types/completion.py b/litellm/types/completion.py
index 1e6cfc0ee33..156ae96410b 100644
--- a/litellm/types/completion.py
+++ b/litellm/types/completion.py
@@ -4,9 +4,11 @@ from collections.abc import Callable, Coroutine, Iterable, Mapping
 from dataclasses import dataclass
 from typing import TYPE_CHECKING, Any, Literal, Union
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict, Field
 from typing_extensions import Required, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 if TYPE_CHECKING:
     import httpx
     from aiohttp import ClientSession
@@ -172,9 +174,9 @@ ChatCompletionMessageParam = (
 )
 
 
-class CompletionRequest(BaseModel):
+class CompletionRequest(LiteLLMBaseModel):
     model: str
-    messages: list[ChatCompletionMessageParam] = []
+    messages: list[ChatCompletionMessageParam] = Field(default=[])
     timeout: float | int | None = None
     temperature: float | None = None
     top_p: float | None = None
diff --git a/litellm/types/containers/main.py b/litellm/types/containers/main.py
index 62ef524a435..afa07462926 100644
--- a/litellm/types/containers/main.py
+++ b/litellm/types/containers/main.py
@@ -2,18 +2,20 @@ import builtins
 from collections.abc import Mapping
 from typing import Any, Literal
 
-from pydantic import BaseModel
+from pydantic import PrivateAttr
 from typing_extensions import ReadOnly, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
 
-class ExpiresAfter(BaseModel):
+
+class ExpiresAfter(LiteLLMBaseModel):
     """Container expiration settings."""
 
     anchor: Literal["last_active_at"]
     minutes: int
 
 
-class ContainerObject(BaseModel):
+class ContainerObject(LiteLLMBaseModel):
     """Represents a container object."""
 
     id: str
@@ -23,7 +25,7 @@ class ContainerObject(BaseModel):
     expires_after: ExpiresAfter | None = None
     last_active_at: int | None = None
     name: str | None = None
-    _hidden_params: dict[str, Any] = {}
+    _hidden_params: dict[str, Any] = PrivateAttr(default={})
 
     def __contains__(self, key: str) -> bool:
         # Define custom behavior for the 'in' operator
@@ -45,7 +47,7 @@ class ContainerObject(BaseModel):
             return self.dict()
 
 
-class DeleteContainerResult(BaseModel):
+class DeleteContainerResult(LiteLLMBaseModel):
     """Result of a delete container request."""
 
     id: str
@@ -68,7 +70,7 @@ class DeleteContainerResult(BaseModel):
             return self.dict()
 
 
-class ContainerListResponse(BaseModel):
+class ContainerListResponse(LiteLLMBaseModel):
     """Response object for list containers request."""
 
     object: Literal["list"]
@@ -130,7 +132,7 @@ class ContainerListOptionalRequestParams(TypedDict, total=False):
     extra_query: dict[str, str] | None
 
 
-class ContainerFileObject(BaseModel):
+class ContainerFileObject(LiteLLMBaseModel):
     """Represents a container file object."""
 
     id: str
@@ -140,7 +142,7 @@ class ContainerFileObject(BaseModel):
     created_at: int
     path: str
     source: str
-    _hidden_params: dict[str, builtins.object] = {}
+    _hidden_params: dict[str, builtins.object] = PrivateAttr(default={})
 
     def __contains__(self, key: str) -> bool:
         return hasattr(self, key)
@@ -158,7 +160,7 @@ class ContainerFileObject(BaseModel):
             return self.dict()
 
 
-class ContainerFileListResponse(BaseModel):
+class ContainerFileListResponse(LiteLLMBaseModel):
     """Response object for list container files request."""
 
     object: Literal["list"]
@@ -183,7 +185,7 @@ class ContainerFileListResponse(BaseModel):
             return self.dict()
 
 
-class DeleteContainerFileResponse(BaseModel):
+class DeleteContainerFileResponse(LiteLLMBaseModel):
     """Response object for delete container file request."""
 
     id: str
diff --git a/litellm/types/embedding.py b/litellm/types/embedding.py
index cc1f518f20b..27a9903fe78 100644
--- a/litellm/types/embedding.py
+++ b/litellm/types/embedding.py
@@ -1,9 +1,11 @@
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict, Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class EmbeddingRequest(BaseModel):
+class EmbeddingRequest(LiteLLMBaseModel):
     model: str
-    input: list[str] = []
+    input: list[str] = Field(default=[])
     timeout: int = 600
     api_base: str | None = None
     api_version: str | None = None
diff --git a/litellm/types/guardrails.py b/litellm/types/guardrails.py
index 8f1140cef62..0c7695d0852 100644
--- a/litellm/types/guardrails.py
+++ b/litellm/types/guardrails.py
@@ -4,10 +4,11 @@ from enum import Enum
 from types import MappingProxyType
 from typing import Final, Literal
 
-from pydantic import BaseModel, ConfigDict, Field, field_validator, model_validator
+from pydantic import ConfigDict, Field, field_validator, model_validator
 from typing_extensions import ReadOnly, Required, TypedDict
 
 from litellm.constants import BEDROCK_APPLY_GUARDRAIL_CHUNK_BUDGET_CHARS
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.guardrails.guardrail_hooks.agent_365 import (
     Agent365GuardrailConfigModel,
 )
@@ -166,7 +167,7 @@ class GuardrailItemSpec(TypedDict, total=False):
     callback_args: dict[str, dict]
 
 
-class GuardrailItem(BaseModel):
+class GuardrailItem(LiteLLMBaseModel):
     callbacks: list[str]
     default_on: bool
     logging_only: bool | None
@@ -436,7 +437,7 @@ class GuardrailParamUITypes(str, Enum):
     PERCENTAGE = "percentage"
 
 
-class PresidioPresidioConfigModelUserInterface(BaseModel):
+class PresidioPresidioConfigModelUserInterface(LiteLLMBaseModel):
     """Configuration parameters for the Presidio PII masking guardrail on LiteLLM UI"""
 
     presidio_analyzer_api_base: str | None = Field(
@@ -545,31 +546,31 @@ BedrockChecksSensitiveInformationEntity = Literal[
 ]
 
 
-class BedrockChecksContentFilterCategoryItem(BaseModel):
+class BedrockChecksContentFilterCategoryItem(LiteLLMBaseModel):
     category: BedrockChecksContentFilterCategory
 
 
-class BedrockChecksContentFilterModel(BaseModel):
+class BedrockChecksContentFilterModel(LiteLLMBaseModel):
     categories: list[BedrockChecksContentFilterCategoryItem]
 
 
-class BedrockChecksPromptAttackCategoryItem(BaseModel):
+class BedrockChecksPromptAttackCategoryItem(LiteLLMBaseModel):
     category: BedrockChecksPromptAttackCategory
 
 
-class BedrockChecksPromptAttackModel(BaseModel):
+class BedrockChecksPromptAttackModel(LiteLLMBaseModel):
     categories: list[BedrockChecksPromptAttackCategoryItem]
 
 
-class BedrockChecksSensitiveInformationEntityItem(BaseModel):
+class BedrockChecksSensitiveInformationEntityItem(LiteLLMBaseModel):
     type: BedrockChecksSensitiveInformationEntity
 
 
-class BedrockChecksSensitiveInformationModel(BaseModel):
+class BedrockChecksSensitiveInformationModel(LiteLLMBaseModel):
     entities: list[BedrockChecksSensitiveInformationEntityItem]
 
 
-class BedrockChecksConfigModel(BaseModel):
+class BedrockChecksConfigModel(LiteLLMBaseModel):
     """Inline `checks` config for the resource-less Bedrock InvokeGuardrailChecks API.
 
     Include only the checks you want to run; at least one must be set.
@@ -588,7 +589,7 @@ class BedrockChecksConfigModel(BaseModel):
         return self
 
 
-class BedrockGuardrailConfigModel(BaseModel):
+class BedrockGuardrailConfigModel(LiteLLMBaseModel):
     """Configuration parameters for the AWS Bedrock guardrail"""
 
     guardrailIdentifier: str | None = Field(default=None, description="The ID of your guardrail on Bedrock")
@@ -663,7 +664,7 @@ class BedrockGuardrailConfigModel(BaseModel):
     )
 
 
-class BedrockGuardrailStreamingParams(BaseModel):
+class BedrockGuardrailStreamingParams(LiteLLMBaseModel):
     streaming_buffer_until_moderated: bool = Field(
         default=True,
         description="If True (default), withhold every streamed chunk until the end-of-stream "
@@ -703,7 +704,7 @@ class BedrockGuardrailStreamingParams(BaseModel):
         )
 
 
-class LakeraV2GuardrailConfigModel(BaseModel):
+class LakeraV2GuardrailConfigModel(LiteLLMBaseModel):
     """Configuration parameters for the Lakera AI v2 guardrail"""
 
     api_key: str | None = Field(default=None, description="API key for the Lakera AI service")
@@ -728,7 +729,7 @@ class LakeraV2GuardrailConfigModel(BaseModel):
     )
 
 
-class LassoGuardrailConfigModel(BaseModel):
+class LassoGuardrailConfigModel(LiteLLMBaseModel):
     """Configuration parameters for the Lasso guardrail"""
 
     lasso_user_id: str | None = Field(default=None, description="User ID for the Lasso guardrail")
@@ -736,7 +737,7 @@ class LassoGuardrailConfigModel(BaseModel):
     mask: bool | None = Field(default=False, description="Enable content masking using Lasso classifix API")
 
 
-class DeepKeepGuardrailConfigModel(BaseModel):
+class DeepKeepGuardrailConfigModel(LiteLLMBaseModel):
     """Configuration parameters for the DeepKeep AI Firewall guardrail"""
 
     deepkeep_firewall_id: str | None = Field(
@@ -748,7 +749,7 @@ class DeepKeepGuardrailConfigModel(BaseModel):
     )
 
 
-class PillarGuardrailConfigModel(BaseModel):
+class PillarGuardrailConfigModel(LiteLLMBaseModel):
     """Configuration parameters for the Pillar Security guardrail"""
 
     on_flagged_action: str | None = Field(
@@ -773,7 +774,7 @@ class PillarGuardrailConfigModel(BaseModel):
     )
 
 
-class NomaGuardrailConfigModel(BaseModel):
+class NomaGuardrailConfigModel(LiteLLMBaseModel):
     """Configuration parameters for the Noma Security guardrail"""
 
     use_v2: bool | None = Field(
@@ -802,7 +803,7 @@ class NomaGuardrailConfigModel(BaseModel):
     )
 
 
-class ZscalerAIGuardConfigModel(BaseModel):
+class ZscalerAIGuardConfigModel(LiteLLMBaseModel):
     """Configuration parameters for the Zscaler AI Guard guardrail"""
 
     policy_id: int | None = Field(
@@ -820,7 +821,7 @@ class ZscalerAIGuardConfigModel(BaseModel):
     )
 
 
-class JavelinGuardrailConfigModel(BaseModel):
+class JavelinGuardrailConfigModel(LiteLLMBaseModel):
     """Configuration parameters for the Javelin guardrail"""
 
     guard_name: str | None = Field(default=None, description="Name of the Javelin guard to use")
@@ -837,7 +838,7 @@ class ContentFilterAction(str, Enum):
     MASK = "MASK"
 
 
-class BlockedWord(BaseModel):
+class BlockedWord(LiteLLMBaseModel):
     """Represents a blocked word with its action and optional description"""
 
     keyword: str = Field(description="The keyword to block or mask")
@@ -848,7 +849,7 @@ class BlockedWord(BaseModel):
     )
 
 
-class ContentFilterPattern(BaseModel):
+class ContentFilterPattern(LiteLLMBaseModel):
     """Represents a content filter pattern (prebuilt or custom regex)"""
 
     pattern_type: Literal["prebuilt", "regex"] = Field(
@@ -869,7 +870,7 @@ class ContentFilterPattern(BaseModel):
     action: ContentFilterAction = Field(description="Action to take when pattern matches (BLOCK or MASK)")
 
 
-class ContentFilterConfigModel(BaseModel):
+class ContentFilterConfigModel(LiteLLMBaseModel):
     """Configuration parameters for the content filter guardrail"""
 
     patterns: list[ContentFilterPattern] | None = Field(
@@ -1168,7 +1169,7 @@ class BaseLitellmParams(ContentFilterConfigModel):  # works for new and patch up
     model_config = ConfigDict(extra="allow", protected_namespaces=())
 
 
-class Mode(BaseModel):
+class Mode(LiteLLMBaseModel):
     tags: dict[str, str | list[str]] = Field(description="Tags for the guardrail mode")
     default: str | list[str] | None = Field(default=None, description="Default mode when no tags match")
 
@@ -1287,7 +1288,7 @@ class GUARDRAIL_DEFINITION_LOCATION(str, Enum):
     CONFIG = "config"
 
 
-class GuardrailInfoResponse(BaseModel):
+class GuardrailInfoResponse(LiteLLMBaseModel):
     guardrail_id: str | None = None
     guardrail_name: str
     litellm_params: BaseLitellmParams | None = None
@@ -1300,11 +1301,11 @@ class GuardrailInfoResponse(BaseModel):
         super().__init__(**kwargs)
 
 
-class ListGuardrailsResponse(BaseModel):
+class ListGuardrailsResponse(LiteLLMBaseModel):
     guardrails: list[GuardrailInfoResponse]
 
 
-class GuardrailUIAddGuardrailSettings(BaseModel):
+class GuardrailUIAddGuardrailSettings(LiteLLMBaseModel):
     supported_entities: list[str]
     supported_actions: list[str]
     supported_modes: list[str]
@@ -1313,7 +1314,7 @@ class GuardrailUIAddGuardrailSettings(BaseModel):
     content_filter_settings: dict[str, object] | None = None
 
 
-class PresidioPerRequestConfig(BaseModel):
+class PresidioPerRequestConfig(LiteLLMBaseModel):
     """
     presdio params that can be controlled per request, api key
     """
@@ -1322,7 +1323,7 @@ class PresidioPerRequestConfig(BaseModel):
     entities: list[PiiEntityType] | None = None
 
 
-class ApplyGuardrailRequest(BaseModel):
+class ApplyGuardrailRequest(LiteLLMBaseModel):
     guardrail_name: str
     text: str
     language: str | None = None
@@ -1332,11 +1333,11 @@ class ApplyGuardrailRequest(BaseModel):
     metadata: dict[str, object] | None = None
 
 
-class ApplyGuardrailResponse(BaseModel):
+class ApplyGuardrailResponse(LiteLLMBaseModel):
     response_text: str
 
 
-class PatchGuardrailRequest(BaseModel):
+class PatchGuardrailRequest(LiteLLMBaseModel):
     guardrail_name: str | None = None
     litellm_params: BaseLitellmParams | None = None
     guardrail_info: dict[str, object] | None = None
diff --git a/litellm/types/integrations/arize.py b/litellm/types/integrations/arize.py
index 7ed7b79e8d1..bde3956683b 100644
--- a/litellm/types/integrations/arize.py
+++ b/litellm/types/integrations/arize.py
@@ -1,6 +1,6 @@
 from typing import TYPE_CHECKING, Any, Literal
 
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     Protocol = Literal["otlp_grpc", "otlp_http"]
@@ -8,7 +8,7 @@ else:
     Protocol = Any
 
 
-class ArizeConfig(BaseModel):
+class ArizeConfig(LiteLLMBaseModel):
     space_id: str | None = None
     space_key: str | None = None
     api_key: str | None = None
diff --git a/litellm/types/integrations/arize_phoenix.py b/litellm/types/integrations/arize_phoenix.py
index ae91c9945b9..05bff51c48f 100644
--- a/litellm/types/integrations/arize_phoenix.py
+++ b/litellm/types/integrations/arize_phoenix.py
@@ -1,9 +1,9 @@
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .arize import Protocol
 
 
-class ArizePhoenixConfig(BaseModel):
+class ArizePhoenixConfig(LiteLLMBaseModel):
     otlp_auth_headers: str | None = None
     protocol: Protocol
     endpoint: str
diff --git a/litellm/types/integrations/custom_logger.py b/litellm/types/integrations/custom_logger.py
index 9a9f3ae34ce..d2db9357349 100644
--- a/litellm/types/integrations/custom_logger.py
+++ b/litellm/types/integrations/custom_logger.py
@@ -1,9 +1,10 @@
 from collections.abc import Mapping
 from typing import Any, Final
 
-from pydantic import BaseModel, Field
+from pydantic import Field
 
 from litellm.types.litellm_params import AgenticSurface
+from litellm.types.llms.base import LiteLLMBaseModel
 
 CHAT_COMPLETION_AGENTIC_SURFACE: Final[AgenticSurface] = "chat_completions"
 RESPONSES_AGENTIC_SURFACE: Final[AgenticSurface] = "responses"
@@ -54,7 +55,7 @@ class AgenticLoopSafetyError(ValueError):
     """
 
 
-class StandardCustomLoggerInitParams(BaseModel):
+class StandardCustomLoggerInitParams(LiteLLMBaseModel):
     """
     Params for initializing a CustomLogger.
     """
@@ -62,7 +63,7 @@ class StandardCustomLoggerInitParams(BaseModel):
     turn_off_message_logging: bool | None = False
 
 
-class AgenticLoopRequestPatch(BaseModel):
+class AgenticLoopRequestPatch(LiteLLMBaseModel):
     """
     Patch returned by callbacks to request a follow-up LLM call.
     """
@@ -75,7 +76,7 @@ class AgenticLoopRequestPatch(BaseModel):
     kwargs: dict[str, Any] = Field(default_factory=dict)
 
 
-class AgenticLoopPlan(BaseModel):
+class AgenticLoopPlan(LiteLLMBaseModel):
     """
     Typed callback response for agentic-loop reruns.
     """
diff --git a/litellm/types/integrations/langfuse_otel.py b/litellm/types/integrations/langfuse_otel.py
index c58dc567cda..4bcd0448c3a 100644
--- a/litellm/types/integrations/langfuse_otel.py
+++ b/litellm/types/integrations/langfuse_otel.py
@@ -1,7 +1,7 @@
 from enum import Enum
 from typing import TYPE_CHECKING, Any, Literal
 
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 if TYPE_CHECKING:
     Protocol = Literal["otlp_grpc", "otlp_http"]
@@ -9,7 +9,7 @@ else:
     Protocol = Any
 
 
-class LangfuseOtelConfig(BaseModel):
+class LangfuseOtelConfig(LiteLLMBaseModel):
     otlp_auth_headers: str | None = None
     protocol: Protocol = "otlp_http"
 
diff --git a/litellm/types/integrations/langsmith.py b/litellm/types/integrations/langsmith.py
index 17fadfec54d..594190ddb5c 100644
--- a/litellm/types/integrations/langsmith.py
+++ b/litellm/types/integrations/langsmith.py
@@ -2,11 +2,12 @@ from dataclasses import dataclass
 from datetime import datetime
 from typing import Any, NamedTuple
 
-from pydantic import BaseModel
 from typing_extensions import TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
 
-class LangsmithInputs(BaseModel):
+
+class LangsmithInputs(LiteLLMBaseModel):
     model: str | None = None
     messages: list[Any] | None = None
     stream: bool | None = None
diff --git a/litellm/types/integrations/s3_v2.py b/litellm/types/integrations/s3_v2.py
index e8ad28f1a3b..c0f719aacfb 100644
--- a/litellm/types/integrations/s3_v2.py
+++ b/litellm/types/integrations/s3_v2.py
@@ -1,11 +1,11 @@
 from typing import Literal
 
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 S3PartitionGranularity = Literal["day", "hour"]
 
 
-class s3BatchLoggingElement(BaseModel):
+class s3BatchLoggingElement(LiteLLMBaseModel):
     """
     Type of element stored in self.log_queue in S3Logger
     """
diff --git a/litellm/types/integrations/slack_alerting.py b/litellm/types/integrations/slack_alerting.py
index 770746196e4..4ba7e277cb6 100644
--- a/litellm/types/integrations/slack_alerting.py
+++ b/litellm/types/integrations/slack_alerting.py
@@ -5,9 +5,10 @@ from datetime import datetime as dt
 from enum import Enum
 from typing import Any, Final, Literal, Optional, Union
 
-from pydantic import BaseModel, Field
+from pydantic import Field
 from typing_extensions import NotRequired, ReadOnly, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import LiteLLMPydanticObjectBase
 
 DEFAULT_DIGEST_INTERVAL: Final = 86400  # 24 hours in seconds
@@ -255,7 +256,7 @@ class AlertQueueItem(TypedDict):
     format: NotRequired[ReadOnly[str]]
 
 
-class HangingRequestData(BaseModel):
+class HangingRequestData(LiteLLMBaseModel):
     request_id: str
     model: str
     api_base: str | None = None
diff --git a/litellm/types/integrations/weave_otel.py b/litellm/types/integrations/weave_otel.py
index d3cf489435f..d99ec7dc51a 100644
--- a/litellm/types/integrations/weave_otel.py
+++ b/litellm/types/integrations/weave_otel.py
@@ -3,10 +3,10 @@ from __future__ import annotations
 from enum import Enum
 from typing import Literal
 
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class WeaveOtelConfig(BaseModel):
+class WeaveOtelConfig(LiteLLMBaseModel):
     """Configuration for Weave OpenTelemetry integration."""
 
     otlp_auth_headers: str | None = None
diff --git a/litellm/types/integrations/websearch_interception.py b/litellm/types/integrations/websearch_interception.py
index 22233404fb3..0ba6e63264a 100644
--- a/litellm/types/integrations/websearch_interception.py
+++ b/litellm/types/integrations/websearch_interception.py
@@ -5,20 +5,21 @@ Type definitions for WebSearch Interception integration.
 from dataclasses import dataclass
 from typing import TYPE_CHECKING, Literal, TypeAlias, TypedDict
 
-from pydantic import BaseModel
 from typing_extensions import ReadOnly
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 if TYPE_CHECKING:
     from litellm.llms.base_llm.search.transformation import SearchResponse
 
 
-class AnthropicSearchQuery(BaseModel):
+class AnthropicSearchQuery(LiteLLMBaseModel):
     """``input`` of an Anthropic ``server_tool_use`` block for a web search."""
 
     query: str
 
 
-class AnthropicServerToolUseBlock(BaseModel):
+class AnthropicServerToolUseBlock(LiteLLMBaseModel):
     """
     The ``server_tool_use`` block that must accompany a ``web_search_tool_result``.
 
diff --git a/litellm/types/interactions/generated.py b/litellm/types/interactions/generated.py
index a666d236e4b..fada4d5728a 100644
--- a/litellm/types/interactions/generated.py
+++ b/litellm/types/interactions/generated.py
@@ -7,10 +7,12 @@ from __future__ import annotations
 from enum import Enum
 from typing import Any, Final, Literal
 
-from pydantic import AwareDatetime, Base64Str, BaseModel, Field, RootModel
+from pydantic import AwareDatetime, Base64Str, Field, RootModel
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class Annotation(BaseModel):
+class Annotation(LiteLLMBaseModel):
     start_index: int | None = Field(
         None,
         description="Start of segment of the response that is attributed to this source.\n\nIndex indicates the start of the segment, measured in bytes.",
@@ -22,14 +24,14 @@ class Annotation(BaseModel):
     )
 
 
-class DocumentContent(BaseModel):
+class DocumentContent(LiteLLMBaseModel):
     data: Base64Str | None = None
     uri: str | None = None
     mime_type: str | None = None
     type: Literal["document"] = Field(..., description="Used as the OpenAPI type discriminator for the content oneof.")
 
 
-class FunctionCallContent(BaseModel):
+class FunctionCallContent(LiteLLMBaseModel):
     name: str = Field(..., description="The name of the tool to call.")
     arguments: dict[str, Any] = Field(..., description="The arguments to pass to the function.")
     type: Literal["function_call"] = Field(
@@ -42,16 +44,16 @@ class Language(Enum):
     python = "python"
 
 
-class CodeExecutionCallArguments(BaseModel):
+class CodeExecutionCallArguments(LiteLLMBaseModel):
     language: Language | None = Field(None, description="Programming language of the `code`.")
     code: str | None = Field(None, description="The code to be executed.")
 
 
-class UrlContextCallArguments(BaseModel):
+class UrlContextCallArguments(LiteLLMBaseModel):
     urls: list[str] | None = Field(None, description="The URLs to fetch.")
 
 
-class McpServerToolCallContent(BaseModel):
+class McpServerToolCallContent(LiteLLMBaseModel):
     name: str = Field(..., description="The name of the tool which was called.")
     server_name: str = Field(..., description="The name of the used MCP server.")
     arguments: dict[str, Any] = Field(..., description="The JSON object of arguments for the function.")
@@ -61,11 +63,11 @@ class McpServerToolCallContent(BaseModel):
     id: str = Field(..., description="A unique ID for this specific tool call.")
 
 
-class GoogleSearchCallArguments(BaseModel):
+class GoogleSearchCallArguments(LiteLLMBaseModel):
     queries: list[str] | None = Field(None, description="Web search queries for the following-up web search.")
 
 
-class CodeExecutionResultContent(BaseModel):
+class CodeExecutionResultContent(LiteLLMBaseModel):
     result: str | None = Field(None, description="The output of the code execution.")
     is_error: bool | None = Field(None, description="Whether the code execution resulted in an error.")
     signature: str | None = Field(None, description="A signature hash for backend validation.")
@@ -82,12 +84,12 @@ class Status(Enum):
     unsafe = "unsafe"
 
 
-class UrlContextResult(BaseModel):
+class UrlContextResult(LiteLLMBaseModel):
     url: str | None = Field(None, description="The URL that was fetched.")
     status: Status | None = Field(None, description="The status of the URL retrieval.")
 
 
-class GoogleSearchResult(BaseModel):
+class GoogleSearchResult(LiteLLMBaseModel):
     url: str | None = Field(None, description="URI reference of the search result.")
     title: str | None = Field(None, description="Title of the search result.")
     rendered_content: str | None = Field(
@@ -96,13 +98,13 @@ class GoogleSearchResult(BaseModel):
     )
 
 
-class FileSearchResult(BaseModel):
+class FileSearchResult(LiteLLMBaseModel):
     title: str | None = Field(None, description="The title of the search result.")
     text: str | None = Field(None, description="The text of the search result.")
     file_search_store: str | None = Field(None, description="The name of the file search store.")
 
 
-class SpeechConfig(BaseModel):
+class SpeechConfig(LiteLLMBaseModel):
     voice: str | None = Field(None, description="The voice of the speaker.")
     language: str | None = Field(None, description="The language of the speech.")
     speaker: str | None = Field(
@@ -111,25 +113,25 @@ class SpeechConfig(BaseModel):
     )
 
 
-class DynamicAgentConfig(BaseModel):
+class DynamicAgentConfig(LiteLLMBaseModel):
     type: Literal["dynamic"] = Field(
         "dynamic",
         description="Used as the OpenAPI type discriminator for the content oneof.",
     )
 
 
-class Function(BaseModel):
+class Function(LiteLLMBaseModel):
     name: str | None = Field(None, description="The name of the function.")
     description: str | None = Field(None, description="A description of the function.")
     parameters: Any | None = Field(None, description="The JSON Schema for the function's parameters.")
     type: Literal["function"]
 
 
-class CodeExecution(BaseModel):
+class CodeExecution(LiteLLMBaseModel):
     type: Literal["code_execution"]
 
 
-class UrlContext(BaseModel):
+class UrlContext(LiteLLMBaseModel):
     type: Literal["url_context"]
 
 
@@ -137,7 +139,7 @@ class Environment(Enum):
     browser = "browser"
 
 
-class ComputerUse(BaseModel):
+class ComputerUse(LiteLLMBaseModel):
     type: Literal["computer_use"]
     environment: Environment | None = Field(None, description="The environment being operated.")
     excludedPredefinedFunctions: list[str] | None = Field(
@@ -146,11 +148,11 @@ class ComputerUse(BaseModel):
     )
 
 
-class GoogleSearch(BaseModel):
+class GoogleSearch(LiteLLMBaseModel):
     type: Literal["google_search"]
 
 
-class FileSearch(BaseModel):
+class FileSearch(LiteLLMBaseModel):
     file_search_store_names: list[str] | None = Field(None, description="The file search store names to search.")
     top_k: int | None = Field(None, description="The number of semantic retrieval chunks to retrieve.")
     metadata_filter: str | None = Field(
@@ -176,7 +178,7 @@ class Status1(Enum):
     queued = "queued"
 
 
-class InteractionStatusUpdate(BaseModel):
+class InteractionStatusUpdate(LiteLLMBaseModel):
     interaction_id: str | None = None
     status: Status1 | None = None
     event_type: Literal["interaction.status_update"] = "interaction.status_update"
@@ -186,20 +188,20 @@ class InteractionStatusUpdate(BaseModel):
     )
 
 
-class TextDelta(BaseModel):
+class TextDelta(LiteLLMBaseModel):
     text: str | None = None
     type: Literal["text"] = Field(..., description="Used as the OpenAPI type discriminator for the content oneof.")
     annotations: list[Annotation] | None = Field(None, description="Citation information for model-generated content.")
 
 
-class DocumentDelta(BaseModel):
+class DocumentDelta(LiteLLMBaseModel):
     data: Base64Str | None = None
     uri: str | None = None
     mime_type: str | None = None
     type: Literal["document"] = Field(..., description="Used as the OpenAPI type discriminator for the content oneof.")
 
 
-class ThoughtSignatureDelta(BaseModel):
+class ThoughtSignatureDelta(LiteLLMBaseModel):
     signature: Base64Str | None = Field(
         None,
         description="Signature to match the backend source to be part of the generation.",
@@ -209,7 +211,7 @@ class ThoughtSignatureDelta(BaseModel):
     )
 
 
-class FunctionCallDelta(BaseModel):
+class FunctionCallDelta(LiteLLMBaseModel):
     name: str | None = None
     arguments: dict[str, Any] | None = None
     type: Literal["function_call"] = Field(
@@ -218,7 +220,7 @@ class FunctionCallDelta(BaseModel):
     id: str | None = Field(None, description="A unique ID for this specific tool call.")
 
 
-class CodeExecutionCallDelta(BaseModel):
+class CodeExecutionCallDelta(LiteLLMBaseModel):
     arguments: CodeExecutionCallArguments | None = None
     type: Literal["code_execution_call"] = Field(
         ..., description="Used as the OpenAPI type discriminator for the content oneof."
@@ -226,7 +228,7 @@ class CodeExecutionCallDelta(BaseModel):
     id: str | None = Field(None, description="A unique ID for this specific tool call.")
 
 
-class UrlContextCallDelta(BaseModel):
+class UrlContextCallDelta(LiteLLMBaseModel):
     arguments: UrlContextCallArguments | None = None
     type: Literal["url_context_call"] = Field(
         ..., description="Used as the OpenAPI type discriminator for the content oneof."
@@ -234,7 +236,7 @@ class UrlContextCallDelta(BaseModel):
     id: str | None = Field(None, description="A unique ID for this specific tool call.")
 
 
-class GoogleSearchCallDelta(BaseModel):
+class GoogleSearchCallDelta(LiteLLMBaseModel):
     arguments: GoogleSearchCallArguments | None = None
     type: Literal["google_search_call"] = Field(
         ..., description="Used as the OpenAPI type discriminator for the content oneof."
@@ -242,7 +244,7 @@ class GoogleSearchCallDelta(BaseModel):
     id: str | None = Field(None, description="A unique ID for this specific tool call.")
 
 
-class McpServerToolCallDelta(BaseModel):
+class McpServerToolCallDelta(LiteLLMBaseModel):
     name: str | None = None
     server_name: str | None = None
     arguments: dict[str, Any] | None = None
@@ -252,7 +254,7 @@ class McpServerToolCallDelta(BaseModel):
     id: str | None = Field(None, description="A unique ID for this specific tool call.")
 
 
-class CodeExecutionResultDelta(BaseModel):
+class CodeExecutionResultDelta(LiteLLMBaseModel):
     result: str | None = None
     is_error: bool | None = None
     signature: str | None = None
@@ -262,7 +264,7 @@ class CodeExecutionResultDelta(BaseModel):
     call_id: str | None = Field(None, description="ID to match the ID from the function call block.")
 
 
-class UrlContextResultDelta(BaseModel):
+class UrlContextResultDelta(LiteLLMBaseModel):
     signature: str | None = None
     result: list[UrlContextResult] | None = None
     is_error: bool | None = None
@@ -272,7 +274,7 @@ class UrlContextResultDelta(BaseModel):
     call_id: str | None = Field(None, description="ID to match the ID from the function call block.")
 
 
-class GoogleSearchResultDelta(BaseModel):
+class GoogleSearchResultDelta(LiteLLMBaseModel):
     signature: str | None = None
     result: list[GoogleSearchResult] | None = None
     is_error: bool | None = None
@@ -282,14 +284,14 @@ class GoogleSearchResultDelta(BaseModel):
     call_id: str | None = Field(None, description="ID to match the ID from the function call block.")
 
 
-class FileSearchResultDelta(BaseModel):
+class FileSearchResultDelta(LiteLLMBaseModel):
     result: list[FileSearchResult] | None = None
     type: Literal["file_search_result"] = Field(
         ..., description="Used as the OpenAPI type discriminator for the content oneof."
     )
 
 
-class ContentStop(BaseModel):
+class ContentStop(LiteLLMBaseModel):
     index: int | None = None
     event_type: Literal["content.stop"] = "content.stop"
     event_id: str | None = Field(
@@ -298,7 +300,7 @@ class ContentStop(BaseModel):
     )
 
 
-class Error(BaseModel):
+class Error(LiteLLMBaseModel):
     code: str | None = Field(None, description="A URI that identifies the error type.")
     message: str | None = Field(None, description="A human-readable error message.")
 
@@ -367,13 +369,13 @@ class VideoMimeTypeOption(RootModel[str]):
     root: str = Field(..., description="The mime type of the video.", title="VideoMimeType")
 
 
-class TextContent(BaseModel):
+class TextContent(LiteLLMBaseModel):
     text: str | None = Field(None, description="The text content.")
     type: Literal["text"] = Field(..., description="Used as the OpenAPI type discriminator for the content oneof.")
     annotations: list[Annotation] | None = Field(None, description="Citation information for model-generated content.")
 
 
-class ImageContent(BaseModel):
+class ImageContent(LiteLLMBaseModel):
     data: Base64Str | None = None
     uri: str | None = None
     mime_type: ImageMimeTypeOption | None = None
@@ -381,14 +383,14 @@ class ImageContent(BaseModel):
     resolution: MediaResolution | None = Field(None, description="The resolution of the media.")
 
 
-class AudioContent(BaseModel):
+class AudioContent(LiteLLMBaseModel):
     data: Base64Str | None = None
     uri: str | None = None
     mime_type: AudioMimeTypeOption | None = None
     type: Literal["audio"] = Field(..., description="Used as the OpenAPI type discriminator for the content oneof.")
 
 
-class VideoContent(BaseModel):
+class VideoContent(LiteLLMBaseModel):
     data: Base64Str | None = None
     uri: str | None = None
     mime_type: VideoMimeTypeOption | None = None
@@ -404,7 +406,7 @@ class ThoughtSummary(RootModel[list[ThoughtSummary1]]):
     root: list[ThoughtSummary1] = Field(..., description="A summary of the thought.")
 
 
-class CodeExecutionCallContent(BaseModel):
+class CodeExecutionCallContent(LiteLLMBaseModel):
     arguments: CodeExecutionCallArguments | None = Field(
         None, description="The arguments to pass to the code execution."
     )
@@ -414,7 +416,7 @@ class CodeExecutionCallContent(BaseModel):
     id: str | None = Field(None, description="A unique ID for this specific tool call.")
 
 
-class UrlContextCallContent(BaseModel):
+class UrlContextCallContent(LiteLLMBaseModel):
     arguments: UrlContextCallArguments | None = Field(None, description="The arguments to pass to the URL context.")
     type: Literal["url_context_call"] = Field(
         ..., description="Used as the OpenAPI type discriminator for the content oneof."
@@ -422,7 +424,7 @@ class UrlContextCallContent(BaseModel):
     id: str | None = Field(None, description="A unique ID for this specific tool call.")
 
 
-class GoogleSearchCallContent(BaseModel):
+class GoogleSearchCallContent(LiteLLMBaseModel):
     arguments: GoogleSearchCallArguments | None = Field(None, description="The arguments to pass to Google Search.")
     type: Literal["google_search_call"] = Field(
         ..., description="Used as the OpenAPI type discriminator for the content oneof."
@@ -430,11 +432,11 @@ class GoogleSearchCallContent(BaseModel):
     id: str | None = Field(None, description="A unique ID for this specific tool call.")
 
 
-class Result(BaseModel):
+class Result(LiteLLMBaseModel):
     items: list[str | ImageContent] | None = None
 
 
-class FunctionResultContent(BaseModel):
+class FunctionResultContent(LiteLLMBaseModel):
     name: str | None = Field(None, description="The name of the tool that was called.")
     is_error: bool | None = Field(None, description="Whether the tool call resulted in an error.")
     type: Literal["function_result"] = Field(
@@ -444,7 +446,7 @@ class FunctionResultContent(BaseModel):
     call_id: str = Field(..., description="ID to match the ID from the function call block.")
 
 
-class UrlContextResultContent(BaseModel):
+class UrlContextResultContent(LiteLLMBaseModel):
     signature: str | None = Field(None, description="The signature of the URL context result.")
     result: list[UrlContextResult] | None = Field(None, description="The results of the URL context.")
     is_error: bool | None = Field(None, description="Whether the URL context resulted in an error.")
@@ -454,7 +456,7 @@ class UrlContextResultContent(BaseModel):
     call_id: str | None = Field(None, description="ID to match the ID from the url context call block.")
 
 
-class GoogleSearchResultContent(BaseModel):
+class GoogleSearchResultContent(LiteLLMBaseModel):
     signature: str | None = Field(None, description="The signature of the Google Search result.")
     result: list[GoogleSearchResult] | None = Field(None, description="The results of the Google Search.")
     is_error: bool | None = Field(None, description="Whether the Google Search resulted in an error.")
@@ -464,7 +466,7 @@ class GoogleSearchResultContent(BaseModel):
     call_id: str | None = Field(None, description="ID to match the ID from the google search call block.")
 
 
-class McpServerToolResultContent(BaseModel):
+class McpServerToolResultContent(LiteLLMBaseModel):
     name: str | None = Field(
         None,
         description="Name of the tool which is called for this specific tool call.",
@@ -477,19 +479,19 @@ class McpServerToolResultContent(BaseModel):
     call_id: str = Field(..., description="ID to match the ID from the MCP server tool call block.")
 
 
-class FileSearchResultContent(BaseModel):
+class FileSearchResultContent(LiteLLMBaseModel):
     result: list[FileSearchResult] | None = Field(None, description="The results of the File Search.")
     type: Literal["file_search_result"] = Field(
         ..., description="Used as the OpenAPI type discriminator for the content oneof."
     )
 
 
-class AllowedTools(BaseModel):
+class AllowedTools(LiteLLMBaseModel):
     mode: ToolChoiceType | None = Field(None, description="The mode of the tool choice.")
     tools: list[str] | None = Field(None, description="The names of the allowed tools.")
 
 
-class DeepResearchAgentConfig(BaseModel):
+class DeepResearchAgentConfig(LiteLLMBaseModel):
     type: Literal["deep-research"] = Field(
         "deep-research",
         description="Used as the OpenAPI type discriminator for the content oneof.",
@@ -499,7 +501,7 @@ class DeepResearchAgentConfig(BaseModel):
     )
 
 
-class McpServer(BaseModel):
+class McpServer(LiteLLMBaseModel):
     type: Literal["mcp_server"]
     name: str | None = Field(None, description="The name of the MCPServer.")
     url: str | None = Field(
@@ -513,12 +515,12 @@ class McpServer(BaseModel):
     allowed_tools: list[AllowedTools] | None = Field(None, description="The allowed tools.")
 
 
-class ModalityTokens(BaseModel):
+class ModalityTokens(LiteLLMBaseModel):
     modality: ResponseModality | None = Field(None, description="The modality associated with the token count.")
     tokens: int | None = Field(None, description="Number of tokens for the modality.")
 
 
-class ImageDelta(BaseModel):
+class ImageDelta(LiteLLMBaseModel):
     data: Base64Str | None = None
     uri: str | None = None
     mime_type: ImageMimeTypeOption | None = None
@@ -526,14 +528,14 @@ class ImageDelta(BaseModel):
     resolution: MediaResolution | None = Field(None, description="The resolution of the media.")
 
 
-class AudioDelta(BaseModel):
+class AudioDelta(LiteLLMBaseModel):
     data: Base64Str | None = None
     uri: str | None = None
     mime_type: AudioMimeTypeOption | None = None
     type: Literal["audio"] = Field(..., description="Used as the OpenAPI type discriminator for the content oneof.")
 
 
-class VideoDelta(BaseModel):
+class VideoDelta(LiteLLMBaseModel):
     data: Base64Str | None = None
     uri: str | None = None
     mime_type: VideoMimeTypeOption | None = None
@@ -541,14 +543,14 @@ class VideoDelta(BaseModel):
     resolution: MediaResolution | None = Field(None, description="The resolution of the media.")
 
 
-class ThoughtSummaryDelta(BaseModel):
+class ThoughtSummaryDelta(LiteLLMBaseModel):
     type: Literal["thought_summary"] = Field(
         ..., description="Used as the OpenAPI type discriminator for the content oneof."
     )
     content: TextContent | ImageContent | None = Field(None, discriminator="type")
 
 
-class FunctionResultDelta(BaseModel):
+class FunctionResultDelta(LiteLLMBaseModel):
     name: str | None = None
     is_error: bool | None = None
     type: Literal["function_result"] = Field(
@@ -558,7 +560,7 @@ class FunctionResultDelta(BaseModel):
     call_id: str | None = Field(None, description="ID to match the ID from the function call block.")
 
 
-class McpServerToolResultDelta(BaseModel):
+class McpServerToolResultDelta(LiteLLMBaseModel):
     name: str | None = None
     server_name: str | None = None
     type: Literal["mcp_server_tool_result"] = Field(
@@ -568,7 +570,7 @@ class McpServerToolResultDelta(BaseModel):
     call_id: str | None = Field(None, description="ID to match the ID from the function call block.")
 
 
-class ErrorEvent(BaseModel):
+class ErrorEvent(LiteLLMBaseModel):
     event_type: Literal["error"] = "error"
     error: Error | None = None
     event_id: str | None = Field(
@@ -577,7 +579,7 @@ class ErrorEvent(BaseModel):
     )
 
 
-class ToolChoiceConfig(BaseModel):
+class ToolChoiceConfig(LiteLLMBaseModel):
     allowed_tools: AllowedTools | None = None
 
 
@@ -587,7 +589,7 @@ class Tool(RootModel[Function | GoogleSearch | CodeExecution | UrlContext | Comp
     )
 
 
-class ThoughtContent(BaseModel):
+class ThoughtContent(LiteLLMBaseModel):
     signature: Base64Str | None = Field(
         None,
         description="Signature to match the backend source to be part of the generation.",
@@ -600,7 +602,7 @@ class ToolChoice(RootModel[ToolChoiceType | ToolChoiceConfig]):
     root: ToolChoiceType | ToolChoiceConfig = Field(..., description="The configuration for tool choice.")
 
 
-class Usage(BaseModel):
+class Usage(LiteLLMBaseModel):
     total_input_tokens: int | None = Field(None, description="Number of tokens in the prompt (context).")
     input_tokens_by_modality: list[ModalityTokens] | None = Field(
         None, description="A breakdown of input token usage by modality."
@@ -629,7 +631,7 @@ class Usage(BaseModel):
     )
 
 
-class ContentDelta(BaseModel):
+class ContentDelta(LiteLLMBaseModel):
     index: int | None = None
     event_type: Literal["content.delta"] = "content.delta"
     event_id: str | None = Field(
@@ -701,7 +703,7 @@ class Content(
     ) = Field(..., description="The content of the response.", discriminator="type")
 
 
-class Turn(BaseModel):
+class Turn(LiteLLMBaseModel):
     role: str | None = Field(
         None,
         description="The originator of this turn. Must be user for input or model for\nmodel output.",
@@ -709,7 +711,7 @@ class Turn(BaseModel):
     content: str | list[Content] | None = Field(None, description="The content of the turn.")
 
 
-class GenerationConfig(BaseModel):
+class GenerationConfig(LiteLLMBaseModel):
     temperature: float | None = Field(None, description="Controls the randomness of the output.")
     top_p: float | None = Field(
         None,
@@ -731,7 +733,7 @@ class GenerationConfig(BaseModel):
     speech_config: list[SpeechConfig] | None = Field(None, description="Configuration for speech interaction.")
 
 
-class ContentStart(BaseModel):
+class ContentStart(LiteLLMBaseModel):
     index: int | None = None
     content: Content | None = None
     event_type: Literal["content.start"] = "content.start"
@@ -741,7 +743,7 @@ class ContentStart(BaseModel):
     )
 
 
-class Interaction(BaseModel):
+class Interaction(LiteLLMBaseModel):
     model: ModelOption | None = Field(None, description="The name of the `Model` used for generating the interaction.")
     agent: AgentOption | None = Field(None, description="The name of the `Agent` used for generating the interaction.")
     id: str = Field(
@@ -797,7 +799,7 @@ class Interaction(BaseModel):
     )
 
 
-class CreateModelInteractionParams(BaseModel):
+class CreateModelInteractionParams(LiteLLMBaseModel):
     model: ModelOption = Field(..., description="The name of the `Model` used for generating the interaction.")
     stream: bool | None = Field(None, description="Input only. Whether the interaction will be streamed.")
     store: bool | None = Field(
@@ -848,7 +850,7 @@ class CreateModelInteractionParams(BaseModel):
     )
 
 
-class CreateAgentInteractionParams(BaseModel):
+class CreateAgentInteractionParams(LiteLLMBaseModel):
     agent: AgentOption = Field(..., description="The name of the `Agent` used for generating the interaction.")
     stream: bool | None = Field(None, description="Input only. Whether the interaction will be streamed.")
     store: bool | None = Field(
@@ -898,7 +900,7 @@ class CreateAgentInteractionParams(BaseModel):
     )
 
 
-class InteractionEvent(BaseModel):
+class InteractionEvent(LiteLLMBaseModel):
     event_type: Literal["interaction.start", "interaction.complete"]
     interaction: Interaction | None = None
     event_id: str | None = Field(
@@ -914,7 +916,7 @@ class InteractionEvent(BaseModel):
 # ---------------------------------------------------------------
 
 
-class StepStart(BaseModel):
+class StepStart(LiteLLMBaseModel):
     """Emitted when a new step begins (replaces content.start)."""
 
     event_type: Literal["step.start"] = "step.start"
@@ -929,7 +931,7 @@ class StepStart(BaseModel):
     )
 
 
-class StepDelta(BaseModel):
+class StepDelta(LiteLLMBaseModel):
     """Emitted for incremental step content (replaces content.delta)."""
 
     event_type: Literal["step.delta"] = "step.delta"
@@ -944,7 +946,7 @@ class StepDelta(BaseModel):
     )
 
 
-class StepStop(BaseModel):
+class StepStop(LiteLLMBaseModel):
     """Emitted when a step finishes (replaces content.stop)."""
 
     event_type: Literal["step.stop"] = "step.stop"
@@ -959,7 +961,7 @@ class StepStop(BaseModel):
     )
 
 
-class InteractionCreated(BaseModel):
+class InteractionCreated(LiteLLMBaseModel):
     """Emitted when the interaction is first created (replaces interaction.start)."""
 
     event_type: Literal["interaction.created"] = "interaction.created"
@@ -970,7 +972,7 @@ class InteractionCreated(BaseModel):
     )
 
 
-class InteractionInProgress(BaseModel):
+class InteractionInProgress(LiteLLMBaseModel):
     """Emitted while the interaction is running."""
 
     event_type: Literal["interaction.in_progress"] = "interaction.in_progress"
@@ -981,7 +983,7 @@ class InteractionInProgress(BaseModel):
     )
 
 
-class InteractionCompleted(BaseModel):
+class InteractionCompleted(LiteLLMBaseModel):
     """Emitted when the interaction finishes (replaces interaction.complete)."""
 
     event_type: Literal["interaction.completed"] = "interaction.completed"
@@ -992,7 +994,7 @@ class InteractionCompleted(BaseModel):
     )
 
 
-class InteractionRequiresAction(BaseModel):
+class InteractionRequiresAction(LiteLLMBaseModel):
     """Emitted when the interaction is paused waiting for a tool result."""
 
     event_type: Literal["interaction.requires_action"] = "interaction.requires_action"
diff --git a/litellm/types/lar1.py b/litellm/types/lar1.py
index 5998b665ece..2a5f096ba51 100644
--- a/litellm/types/lar1.py
+++ b/litellm/types/lar1.py
@@ -1,6 +1,8 @@
 from enum import Enum
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
 class LAR1Act(str, Enum):
@@ -31,9 +33,9 @@ class LAR1Evidence(str, Enum):
     CONFIRMED = "CONFIRMED"
 
 
-class LAR1Metadata(BaseModel):
+class LAR1Metadata(LiteLLMBaseModel):
     act: LAR1Act = LAR1Act.INF
     time: LAR1Time = LAR1Time.NOW
     mind: LAR1Mind = LAR1Mind.REF
     confidence: float = Field(default=0.5, ge=0.0, le=1.0)
-    evidence: list[LAR1Evidence] = []
+    evidence: list[LAR1Evidence] = Field(default=[])
diff --git a/litellm/types/llms/anthropic.py b/litellm/types/llms/anthropic.py
index 46693009bfd..f0b5a1bb84e 100644
--- a/litellm/types/llms/anthropic.py
+++ b/litellm/types/llms/anthropic.py
@@ -2,9 +2,11 @@ from collections.abc import Iterable, Sequence
 from enum import Enum
 from typing import Any, Final, Literal, TypeAlias
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 from typing_extensions import NotRequired, ReadOnly, Required, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 from .openai import (
     ChatCompletionCachedContent,
     ChatCompletionRedactedThinkingBlock,
@@ -639,12 +641,12 @@ class MessageStartBlock(TypedDict):
     message: MessageChunk
 
 
-class AnthropicResponseContentBlockText(BaseModel):
+class AnthropicResponseContentBlockText(LiteLLMBaseModel):
     type: Literal["text"]
     text: str
 
 
-class AnthropicResponseContentBlockToolUse(BaseModel):
+class AnthropicResponseContentBlockToolUse(LiteLLMBaseModel):
     type: Literal["tool_use"]
     id: str
     name: str
@@ -654,25 +656,25 @@ class AnthropicResponseContentBlockToolUse(BaseModel):
     model_config = ConfigDict(extra="allow")  # Allow provider_specific_fields
 
 
-class AnthropicResponseContentBlockThinking(BaseModel):
+class AnthropicResponseContentBlockThinking(LiteLLMBaseModel):
     type: Literal["thinking"]
     thinking: str
     signature: str | None
 
 
-class AnthropicResponseContentBlockRedactedThinking(BaseModel):
+class AnthropicResponseContentBlockRedactedThinking(LiteLLMBaseModel):
     type: Literal["redacted_thinking"]
     data: str
 
 
-class AnthropicResponseUsageBlock(BaseModel):
+class AnthropicResponseUsageBlock(LiteLLMBaseModel):
     model_config = ConfigDict(extra="allow")
 
     input_tokens: int
     output_tokens: int
 
 
-class AnthropicOutputTokensDetails(BaseModel):
+class AnthropicOutputTokensDetails(LiteLLMBaseModel):
     model_config = ConfigDict(extra="allow")
 
     thinking_tokens: int | None = None
@@ -681,7 +683,7 @@ class AnthropicOutputTokensDetails(BaseModel):
 AnthropicFinishReason = Literal["end_turn", "max_tokens", "stop_sequence", "tool_use", "refusal"]
 
 
-class AnthropicResponse(BaseModel):
+class AnthropicResponse(LiteLLMBaseModel):
     id: str
     """Unique object identifier."""
 
diff --git a/litellm/types/llms/anthropic_messages/anthropic_request.py b/litellm/types/llms/anthropic_messages/anthropic_request.py
index cbdd1c4446a..d1f66394ca5 100644
--- a/litellm/types/llms/anthropic_messages/anthropic_request.py
+++ b/litellm/types/llms/anthropic_messages/anthropic_request.py
@@ -1,7 +1,7 @@
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class AnthropicMetadata(BaseModel):
+class AnthropicMetadata(LiteLLMBaseModel):
     """
     Object with allowed fields for Anthropic API metadata
 
diff --git a/litellm/types/llms/anthropic_skills.py b/litellm/types/llms/anthropic_skills.py
index 4b27f9b17ef..d11fd32f88b 100644
--- a/litellm/types/llms/anthropic_skills.py
+++ b/litellm/types/llms/anthropic_skills.py
@@ -4,9 +4,10 @@ Type definitions for Anthropic Skills API
 
 from typing import Any
 
-from pydantic import BaseModel
 from typing_extensions import TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 
 # Skills API Request Types
 class CreateSkillRequest(TypedDict, total=False):
@@ -33,7 +34,7 @@ class ListSkillsParams(TypedDict, total=False):
 
 
 # Skills API Response Types
-class Skill(BaseModel):
+class Skill(LiteLLMBaseModel):
     """Represents a skill from the Anthropic Skills API"""
 
     id: str
@@ -67,7 +68,7 @@ class Skill(BaseModel):
     unless a query was given."""
 
 
-class ListSkillsResponse(BaseModel):
+class ListSkillsResponse(LiteLLMBaseModel):
     """Response from listing skills"""
 
     data: list[Skill]
@@ -80,7 +81,7 @@ class ListSkillsResponse(BaseModel):
     """Whether there are more skills available"""
 
 
-class DeleteSkillResponse(BaseModel):
+class DeleteSkillResponse(LiteLLMBaseModel):
     """Response from deleting a skill"""
 
     id: str
@@ -107,7 +108,7 @@ class CreateSkillVersionRequest(TypedDict, total=False):
     """Additional metadata"""
 
 
-class SkillVersion(BaseModel):
+class SkillVersion(LiteLLMBaseModel):
     """Represents a skill version"""
 
     id: str
@@ -135,7 +136,7 @@ class SkillVersion(BaseModel):
     """Object type"""
 
 
-class ListSkillVersionsResponse(BaseModel):
+class ListSkillVersionsResponse(LiteLLMBaseModel):
     """Response from listing skill versions"""
 
     object: str = "list"
@@ -154,7 +155,7 @@ class ListSkillVersionsResponse(BaseModel):
     """Whether there are more versions available"""
 
 
-class DeleteSkillVersionResponse(BaseModel):
+class DeleteSkillVersionResponse(LiteLLMBaseModel):
     """Response from deleting a skill version"""
 
     id: str
diff --git a/litellm/types/llms/base.py b/litellm/types/llms/base.py
index 938aa8064c9..803e7cb44b0 100644
--- a/litellm/types/llms/base.py
+++ b/litellm/types/llms/base.py
@@ -1,10 +1,28 @@
-from typing import Any, Final
+from typing import TYPE_CHECKING, Any, Final
 
 from openai._models import BaseModel as OpenAIObject
 from pydantic import BaseModel, ConfigDict
 
+from litellm.constants import DEFER_PYDANTIC_BUILD
 
-class LiteLLMPydanticObjectBase(BaseModel):
+
+class LiteLLMBaseModel(BaseModel):
+    model_config = ConfigDict(defer_build=DEFER_PYDANTIC_BUILD)
+
+    if TYPE_CHECKING:
+
+        def __init__(self, /, **data: object) -> None: ...
+
+    def model_post_init(self, context: object, /) -> None:
+        # Instances built by a parent's validator or by model_construct skip this class's own
+        # first-use build, leaving a MockValSer that fails when the instance is serialized as Any
+        # TODO: drop once pydantic ships a fix for https://github.com/pydantic/pydantic/issues/13647
+        model = type(self)
+        if not model.__pydantic_complete__:
+            model.model_rebuild(raise_errors=False)
+
+
+class LiteLLMPydanticObjectBase(LiteLLMBaseModel):
     """
     Implements default functions, all pydantic objects should have.
     """
@@ -26,7 +44,7 @@ class LiteLLMPydanticObjectBase(BaseModel):
     model_config = ConfigDict(protected_namespaces=())
 
 
-class BaseLiteLLMOpenAIResponseObject(BaseModel):
+class BaseLiteLLMOpenAIResponseObject(LiteLLMBaseModel):
     model_config = ConfigDict(extra="allow", protected_namespaces=())
 
     def __getitem__(self, key):
@@ -77,7 +95,7 @@ class HiddenParams(OpenAIObject):
         return data
 
 
-class CachedTokensDetails(BaseModel):
+class CachedTokensDetails(LiteLLMBaseModel):
     text_tokens: int | None = None
     audio_tokens: int | None = None
     image_tokens: int | None = None
diff --git a/litellm/types/llms/bedrock.py b/litellm/types/llms/bedrock.py
index e4c41c3ee5b..cc420fb9509 100644
--- a/litellm/types/llms/bedrock.py
+++ b/litellm/types/llms/bedrock.py
@@ -3,9 +3,11 @@ from collections.abc import Sequence
 from enum import Enum
 from typing import TYPE_CHECKING, Any, Final, Literal, TypeAlias
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 from typing_extensions import NotRequired, ReadOnly, Required, TypedDict, override
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 from .openai import ChatCompletionToolCallChunk
 
 
@@ -1114,7 +1116,7 @@ class AwsSessionTag(TypedDict):
     Value: str  # writable-ok: boto3's STS stubs type assume_role Tags as writable TagTypeDef, which rejects ReadOnly
 
 
-class AwsAuthParams(BaseModel):
+class AwsAuthParams(LiteLLMBaseModel):
     """Every credential-shaped aws_* param BaseAWSLLM.get_credentials accepts; region is resolved separately."""
 
     model_config = ConfigDict(frozen=True, extra="ignore")
diff --git a/litellm/types/llms/gemini.py b/litellm/types/llms/gemini.py
index 57fb8b5b0cd..c300c26bcd3 100644
--- a/litellm/types/llms/gemini.py
+++ b/litellm/types/llms/gemini.py
@@ -151,7 +151,7 @@ class BidiGenerateContentSetup(TypedDict, total=False):
 
 
 # Image Generation Types
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
 class GeminiImageGenerationInstance(TypedDict):
@@ -160,7 +160,7 @@ class GeminiImageGenerationInstance(TypedDict):
     prompt: str
 
 
-class GeminiImageGenerationParameters(BaseModel):
+class GeminiImageGenerationParameters(LiteLLMBaseModel):
     """Parameters for Gemini image generation request"""
 
     sampleCount: int | None = None
@@ -204,7 +204,7 @@ class GeminiImageGenerationParameters(BaseModel):
     """User specification"""
 
 
-class GeminiImageGenerationRequest(BaseModel):
+class GeminiImageGenerationRequest(LiteLLMBaseModel):
     """Complete request body for Gemini image generation"""
 
     instances: list[GeminiImageGenerationInstance]
@@ -238,7 +238,7 @@ class GeminiVideoGenerationInstance(TypedDict, total=False):
     image: dict[str, Any]
 
 
-class GeminiVideoGenerationParameters(BaseModel):
+class GeminiVideoGenerationParameters(LiteLLMBaseModel):
     """
     Parameters for Gemini video generation request.
 
@@ -290,7 +290,7 @@ class GeminiVideoGenerationParameters(BaseModel):
     """
 
 
-class GeminiVideoGenerationRequest(BaseModel):
+class GeminiVideoGenerationRequest(LiteLLMBaseModel):
     """Complete request body for Gemini video generation"""
 
     instances: list[GeminiVideoGenerationInstance]
@@ -298,35 +298,35 @@ class GeminiVideoGenerationRequest(BaseModel):
 
 
 # Video Generation Operation Response Types
-class GeminiVideoUri(BaseModel):
+class GeminiVideoUri(LiteLLMBaseModel):
     """Video URI in the generated sample"""
 
     uri: str
     """File URI of the generated video (e.g., 'files/abc123...')"""
 
 
-class GeminiGeneratedVideoSample(BaseModel):
+class GeminiGeneratedVideoSample(LiteLLMBaseModel):
     """Individual generated video sample"""
 
     video: GeminiVideoUri
     """Video object containing the URI"""
 
 
-class GeminiGenerateVideoResponse(BaseModel):
+class GeminiGenerateVideoResponse(LiteLLMBaseModel):
     """Generate video response containing the samples"""
 
     generatedSamples: list[GeminiGeneratedVideoSample]
     """List of generated video samples"""
 
 
-class GeminiOperationResponse(BaseModel):
+class GeminiOperationResponse(LiteLLMBaseModel):
     """Response object in the operation when done"""
 
     generateVideoResponse: GeminiGenerateVideoResponse
     """Video generation response"""
 
 
-class GeminiOperationMetadata(BaseModel):
+class GeminiOperationMetadata(LiteLLMBaseModel):
     """Metadata for the operation"""
 
     createTime: str | None = None
@@ -335,7 +335,7 @@ class GeminiOperationMetadata(BaseModel):
     """Model used for generation"""
 
 
-class GeminiLongRunningOperationResponse(BaseModel):
+class GeminiLongRunningOperationResponse(LiteLLMBaseModel):
     """
     Complete response for a long-running operation.
 
diff --git a/litellm/types/llms/gemini_audio_transcription.py b/litellm/types/llms/gemini_audio_transcription.py
index f7e74ba4bf8..c1cf3452945 100644
--- a/litellm/types/llms/gemini_audio_transcription.py
+++ b/litellm/types/llms/gemini_audio_transcription.py
@@ -1,8 +1,10 @@
 from typing import Literal
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 from typing_extensions import ReadOnly, Required, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 
 class GeminiTranscriptionAudioInput(TypedDict):
     type: ReadOnly[Literal["audio"]]
@@ -31,7 +33,7 @@ class GeminiTranscriptionInteractionRequest(TypedDict, total=False):
     generation_config: ReadOnly[GeminiTranscriptionGenerationConfig]
 
 
-class GeminiTranscriptionWordAnnotation(BaseModel):
+class GeminiTranscriptionWordAnnotation(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     type: str | None = None
@@ -41,7 +43,7 @@ class GeminiTranscriptionWordAnnotation(BaseModel):
     end_offset: str | None = None
 
 
-class GeminiTranscriptionContent(BaseModel):
+class GeminiTranscriptionContent(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     type: str | None = None
@@ -49,21 +51,21 @@ class GeminiTranscriptionContent(BaseModel):
     annotations: tuple[GeminiTranscriptionWordAnnotation, ...] = ()
 
 
-class GeminiTranscriptionStep(BaseModel):
+class GeminiTranscriptionStep(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     type: str | None = None
     content: tuple[GeminiTranscriptionContent, ...] = ()
 
 
-class GeminiTranscriptionModalityTokens(BaseModel):
+class GeminiTranscriptionModalityTokens(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     modality: str | None = None
     tokens: int = 0
 
 
-class GeminiTranscriptionUsage(BaseModel):
+class GeminiTranscriptionUsage(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     total_tokens: int = 0
@@ -72,7 +74,7 @@ class GeminiTranscriptionUsage(BaseModel):
     input_tokens_by_modality: tuple[GeminiTranscriptionModalityTokens, ...] = ()
 
 
-class GeminiTranscriptionInteractionResponse(BaseModel):
+class GeminiTranscriptionInteractionResponse(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     id: str | None = None
diff --git a/litellm/types/llms/oci.py b/litellm/types/llms/oci.py
index d87e4231337..bd3a0af1dca 100644
--- a/litellm/types/llms/oci.py
+++ b/litellm/types/llms/oci.py
@@ -3,7 +3,9 @@ from __future__ import annotations
 from enum import Enum
 from typing import Literal
 
-from pydantic import BaseModel, SerializeAsAny
+from pydantic import SerializeAsAny
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 OCIRoles = Literal["SYSTEM", "USER", "ASSISTANT", "TOOL"]
 
@@ -21,7 +23,7 @@ class OCIVendors(Enum):
 # --- Base Models and Content Parts ---
 
 
-class OCIContentPart(BaseModel):
+class OCIContentPart(LiteLLMBaseModel):
     """Base model for content parts in an OCI message."""
 
 
@@ -32,7 +34,7 @@ class OCITextContentPart(OCIContentPart):
     text: str
 
 
-class OCIImageUrl(BaseModel):
+class OCIImageUrl(LiteLLMBaseModel):
     """ImageUrl object for OCI API. See: https://docs.oracle.com/en-us/iaas/tools/python/latest/api/generative_ai_inference/models/oci.generative_ai_inference.models.ImageUrl.html"""
 
     url: str
@@ -51,7 +53,7 @@ OCIContentPartUnion = OCITextContentPart | OCIImageContentPart
 # --- Models for Tools and Tool Calls ---
 
 
-class OCIToolCall(BaseModel):
+class OCIToolCall(LiteLLMBaseModel):
     """Represents a tool call made by the model."""
 
     id: str | None = None  # absent in some provider responses (e.g. Google via OCI)
@@ -60,7 +62,7 @@ class OCIToolCall(BaseModel):
     arguments: str  # Arguments should be a JSON-serialized string
 
 
-class OCIToolDefinition(BaseModel):
+class OCIToolDefinition(LiteLLMBaseModel):
     """Defines a tool that can be used by the model."""
 
     type: Literal["FUNCTION"] = "FUNCTION"
@@ -72,7 +74,7 @@ class OCIToolDefinition(BaseModel):
 # --- Message Models (Request and Response) ---
 
 
-class OCIMessage(BaseModel):
+class OCIMessage(LiteLLMBaseModel):
     """Model for a single message in the request/response payload."""
 
     role: OCIRoles
@@ -84,7 +86,7 @@ class OCIMessage(BaseModel):
 # --- Request Payload Models ---
 
 
-class OCIChatRequestPayload(BaseModel):
+class OCIChatRequestPayload(LiteLLMBaseModel):
     """Internal 'chatRequest' payload for the OCI API."""
 
     apiFormat: str
@@ -111,7 +113,7 @@ class OCIChatRequestPayload(BaseModel):
     logProbs: int | None = None
 
 
-class OCIServingMode(BaseModel):
+class OCIServingMode(LiteLLMBaseModel):
     """Defines the serving mode and the model to be used."""
 
     servingType: str
@@ -119,7 +121,7 @@ class OCIServingMode(BaseModel):
     modelId: str | None = None
 
 
-class OCICompletionPayload(BaseModel):
+class OCICompletionPayload(LiteLLMBaseModel):
     """Pydantic model for the complete OCI chat request body."""
 
     compartmentId: str
@@ -130,20 +132,20 @@ class OCICompletionPayload(BaseModel):
 # --- API Response Models (Non-streaming) ---
 
 
-class OCICompletionTokenDetails(BaseModel):
+class OCICompletionTokenDetails(LiteLLMBaseModel):
     """Completion token details in the OCI response."""
 
     acceptedPredictionTokens: int | None = None
     reasoningTokens: int | None = None
 
 
-class OCIPromptTokensDetails(BaseModel):
+class OCIPromptTokensDetails(LiteLLMBaseModel):
     """Prompt token details in the OCI response."""
 
     cachedTokens: int | None = None
 
 
-class OCIResponseUsage(BaseModel):
+class OCIResponseUsage(LiteLLMBaseModel):
     """Token usage in the OCI response."""
 
     promptTokens: int
@@ -155,7 +157,7 @@ class OCIResponseUsage(BaseModel):
     promptTokensDetails: OCIPromptTokensDetails | None = None
 
 
-class OCIResponseChoice(BaseModel):
+class OCIResponseChoice(LiteLLMBaseModel):
     """A completion choice in the OCI response."""
 
     index: int
@@ -166,7 +168,7 @@ class OCIResponseChoice(BaseModel):
     logprobs: dict[str, object] | None = None
 
 
-class OCIChatResponse(BaseModel):
+class OCIChatResponse(LiteLLMBaseModel):
     """The 'chatResponse' object in the OCI response."""
 
     apiFormat: str
@@ -175,7 +177,7 @@ class OCIChatResponse(BaseModel):
     usage: OCIResponseUsage
 
 
-class OCICompletionResponse(BaseModel):
+class OCICompletionResponse(LiteLLMBaseModel):
     """Model for the complete non-streaming OCI response body."""
 
     modelId: str
@@ -186,7 +188,7 @@ class OCICompletionResponse(BaseModel):
 # --- API Response Models (Streaming) ---
 
 
-class OCIStreamDelta(BaseModel):
+class OCIStreamDelta(LiteLLMBaseModel):
     """The content delta in a streaming chunk."""
 
     content: list[OCIContentPartUnion] | None = None
@@ -194,7 +196,7 @@ class OCIStreamDelta(BaseModel):
     toolCalls: list[OCIToolCall] | None = None
 
 
-class OCIStreamChunk(BaseModel):
+class OCIStreamChunk(LiteLLMBaseModel):
     """Model for a single SSE event chunk from OCI."""
 
     finishReason: str | None = None
@@ -206,7 +208,7 @@ class OCIStreamChunk(BaseModel):
 # --- Cohere-Specific Models ---
 
 
-class CohereStreamChunk(BaseModel):
+class CohereStreamChunk(LiteLLMBaseModel):
     """Model for a single SSE event chunk from OCI Cohere API."""
 
     apiFormat: str
@@ -218,7 +220,7 @@ class CohereStreamChunk(BaseModel):
     index: int | None = None
 
 
-class CohereMessage(BaseModel):
+class CohereMessage(LiteLLMBaseModel):
     """Base model for Cohere messages."""
 
     role: str
@@ -255,7 +257,7 @@ class CohereToolMessage(CohereMessage):
     toolResults: list[CohereToolResult]
 
 
-class CohereParameterDefinition(BaseModel):
+class CohereParameterDefinition(LiteLLMBaseModel):
     """Parameter definition for Cohere tools."""
 
     description: str
@@ -263,7 +265,7 @@ class CohereParameterDefinition(BaseModel):
     isRequired: bool = False
 
 
-class CohereTool(BaseModel):
+class CohereTool(LiteLLMBaseModel):
     """Tool definition for Cohere."""
 
     name: str
@@ -271,14 +273,14 @@ class CohereTool(BaseModel):
     parameterDefinitions: dict[str, CohereParameterDefinition]
 
 
-class CohereToolCall(BaseModel):
+class CohereToolCall(LiteLLMBaseModel):
     """Tool call made by Cohere model."""
 
     name: str
     parameters: dict[str, object]
 
 
-class CohereToolResult(BaseModel):
+class CohereToolResult(LiteLLMBaseModel):
     """Result of a tool call.
 
     Matches the OCI SDK's CohereToolResult: each result carries the originating
@@ -289,7 +291,7 @@ class CohereToolResult(BaseModel):
     outputs: list[dict[str, object]]
 
 
-class CohereChatRequest(BaseModel):
+class CohereChatRequest(LiteLLMBaseModel):
     """Cohere chat request model."""
 
     # Required fields
@@ -336,7 +338,7 @@ class CohereChatRequest(BaseModel):
     streamOptions: dict[str, object] | None = None
 
 
-class CohereUsage(BaseModel):
+class CohereUsage(LiteLLMBaseModel):
     """Usage information for Cohere response."""
 
     promptTokens: int
@@ -346,7 +348,7 @@ class CohereUsage(BaseModel):
     completionTokensDetails: dict[str, object] | None = None
 
 
-class CohereCitation(BaseModel):
+class CohereCitation(LiteLLMBaseModel):
     """Citation in Cohere response."""
 
     start: int
@@ -355,14 +357,14 @@ class CohereCitation(BaseModel):
     document_ids: list[str]
 
 
-class CohereSearchQuery(BaseModel):
+class CohereSearchQuery(LiteLLMBaseModel):
     """Search query generated by Cohere."""
 
     text: str
     generation_id: str
 
 
-class CohereChatResponse(BaseModel):
+class CohereChatResponse(LiteLLMBaseModel):
     """Cohere chat response model."""
 
     # Required fields
@@ -387,7 +389,7 @@ class CohereChatResponse(BaseModel):
     usage: CohereUsage | None = None
 
 
-class CohereChatDetails(BaseModel):
+class CohereChatDetails(LiteLLMBaseModel):
     """Chat details for Cohere request."""
 
     compartmentId: str
@@ -395,7 +397,7 @@ class CohereChatDetails(BaseModel):
     chatRequest: CohereChatRequest
 
 
-class CohereChatResult(BaseModel):
+class CohereChatResult(LiteLLMBaseModel):
     """Complete Cohere chat result."""
 
     modelId: str
@@ -408,7 +410,7 @@ class CohereChatResult(BaseModel):
 # ---------------------------------------------------------------------------
 
 
-class OCIEmbedRequest(BaseModel):
+class OCIEmbedRequest(LiteLLMBaseModel):
     """Request body for POST /20231130/actions/embedText."""
 
     compartmentId: str
@@ -419,12 +421,12 @@ class OCIEmbedRequest(BaseModel):
     outputDimensions: int | None = None  # cohere.embed-v4.0+; valid: 256, 512, 1024, 1536
 
 
-class OCIEmbedUsage(BaseModel):
+class OCIEmbedUsage(LiteLLMBaseModel):
     promptTokens: int
     totalTokens: int
 
 
-class OCIEmbedResponse(BaseModel):
+class OCIEmbedResponse(LiteLLMBaseModel):
     """Response body from POST /20231130/actions/embedText."""
 
     id: str | None = None  # present in the official SDK response
diff --git a/litellm/types/llms/openai.py b/litellm/types/llms/openai.py
index 6db7fd68292..f241c4a7a27 100644
--- a/litellm/types/llms/openai.py
+++ b/litellm/types/llms/openai.py
@@ -63,7 +63,6 @@ from openai.types.responses.response_create_params import (
 from openai.types.responses.response_function_tool_call import ResponseFunctionToolCall
 from openai.types.responses.response_function_web_search import ResponseFunctionWebSearch
 from pydantic import (
-    BaseModel,
     ConfigDict,
     Discriminator,
     Field,
@@ -82,7 +81,7 @@ from typing_extensions import (
     override,
 )
 
-from litellm.types.llms.base import BaseLiteLLMOpenAIResponseObject
+from litellm.types.llms.base import BaseLiteLLMOpenAIResponseObject, LiteLLMBaseModel
 from litellm.types.responses.main import (
     CustomToolCallOutputItem,
     GenericResponseOutputItem,
@@ -283,7 +282,7 @@ class MessageData(TypedDict):
     metadata: dict | None
 
 
-class Thread(BaseModel):
+class Thread(LiteLLMBaseModel):
     id: str
     """The identifier, which can be referenced in API endpoints."""
 
@@ -318,7 +317,7 @@ OpenAIFilesPurpose = Literal[
 ]
 
 
-class BatchGuardrailRecord(BaseModel):
+class BatchGuardrailRecord(LiteLLMBaseModel):
     """One batch input record a guardrail acted on."""
 
     line: int
@@ -342,7 +341,7 @@ class BatchGuardrailRecord(BaseModel):
     """
 
 
-class BatchGuardrailReport(BaseModel):
+class BatchGuardrailReport(LiteLLMBaseModel):
     """What guardrails did to a batch input file, per record."""
 
     submitted_records: int
@@ -359,7 +358,7 @@ _JsonValue: TypeAlias = object
 BATCH_GUARDRAIL_RESPONSE_FIELD: Final = "litellm_batch_guardrail"
 
 
-class OpenAIFileObject(BaseModel):
+class OpenAIFileObject(LiteLLMBaseModel):
     id: str
     """The file identifier, which can be referenced in the API endpoints."""
 
@@ -405,7 +404,7 @@ class OpenAIFileObject(BaseModel):
     Absent on every other upload, so OpenAI-shaped clients see an unchanged response.
     """
 
-    _hidden_params: dict = {"response_cost": 0.0}  # no cost for writing a file
+    _hidden_params: dict = PrivateAttr(default={"response_cost": 0.0})  # no cost for writing a file
 
     @model_serializer(mode="wrap")
     def _omit_absent_batch_guardrail(  # noqa: ANN202  # annotating it replaces the model's serialization schema
@@ -436,7 +435,7 @@ class OpenAIFileObject(BaseModel):
             return self.dict()
 
 
-class FileListPage(BaseModel):
+class FileListPage(LiteLLMBaseModel):
     """A page of files, as `GET /v1/files` returns it.
 
     Post-call hooks and logging callbacks are handed the listing response, and
@@ -1115,7 +1114,7 @@ class OpenAIChatCompletionChunk(ChatCompletionChunk):
         super().__init__(**kwargs)
 
 
-class Hyperparameters(BaseModel):
+class Hyperparameters(LiteLLMBaseModel):
     batch_size: str | int | None = None  # "Number of examples in each batch."
     learning_rate_multiplier: str | float | None = None  # Scaling factor for the learning rate
     n_epochs: str | int | None = None  # "The number of epochs to train the model for"
@@ -1123,7 +1122,7 @@ class Hyperparameters(BaseModel):
     model_config = {"extra": "allow"}
 
 
-class FineTuningJobCreate(BaseModel):
+class FineTuningJobCreate(LiteLLMBaseModel):
     """
     FineTuningJobCreate - Create a fine-tuning job
 
@@ -1157,7 +1156,7 @@ class FineTuningJobCreate(BaseModel):
 class LiteLLMFineTuningJobCreate(FineTuningJobCreate):
     custom_llm_provider: Literal["openai", "azure", "vertex_ai"] | None = None
 
-    model_config = {"extra": "allow"}  # This allows the model to accept additional fields
+    model_config = ConfigDict(extra="allow")  # This allows the model to accept additional fields
 
 
 AllEmbeddingInputValues = str | list[str] | list[int] | list[list[int]]
@@ -1363,16 +1362,16 @@ class ResponseAPIUsage(BaseLiteLLMOpenAIResponseObject):
             return v.get("total_cost")
         return v
 
-    model_config = {"extra": "allow"}
+    model_config = ConfigDict(extra="allow")
 
 
-class WebSearchToolUsage(BaseModel):
+class WebSearchToolUsage(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     num_requests: NonNegativeInt
 
 
-class ResponsesToolUsage(BaseModel):
+class ResponsesToolUsage(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     web_search: WebSearchToolUsage | None = None
@@ -2519,7 +2518,7 @@ class CreateVideoRequest(TypedDict, total=False):
     timeout: float | None
 
 
-class OpenAIVideoObject(BaseModel):
+class OpenAIVideoObject(LiteLLMBaseModel):
     """OpenAI Video Object representing a video generation job."""
 
     id: str
@@ -2558,7 +2557,7 @@ class OpenAIVideoObject(BaseModel):
     model: str | None = None
     """The video generation model that produced the job."""
 
-    _hidden_params: dict[str, _JsonValue] = {}
+    _hidden_params: dict[str, _JsonValue] = PrivateAttr(default={})
 
     def __contains__(self, key) -> bool:
         return hasattr(self, key)
diff --git a/litellm/types/llms/openai_evals.py b/litellm/types/llms/openai_evals.py
index 519e3e82fff..b753d17e719 100644
--- a/litellm/types/llms/openai_evals.py
+++ b/litellm/types/llms/openai_evals.py
@@ -5,9 +5,10 @@ Type definitions for OpenAI Evals API
 import builtins
 from typing import Literal
 
-from pydantic import BaseModel
 from typing_extensions import Required, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 
 # Evals API Request Types
 class DataSourceConfigCustom(TypedDict, total=False):
@@ -128,7 +129,7 @@ class ListEvalsParams(TypedDict, total=False):
 
 
 # Evals API Response Types
-class Eval(BaseModel):
+class Eval(LiteLLMBaseModel):
     """Represents an evaluation from the OpenAI Evals API"""
 
     id: str
@@ -156,7 +157,7 @@ class Eval(BaseModel):
     """Additional metadata"""
 
 
-class ListEvalsResponse(BaseModel):
+class ListEvalsResponse(LiteLLMBaseModel):
     """Response from listing evaluations"""
 
     object: str = "list"
@@ -175,7 +176,7 @@ class ListEvalsResponse(BaseModel):
     """Whether there are more evaluations available"""
 
 
-class DeleteEvalResponse(BaseModel):
+class DeleteEvalResponse(LiteLLMBaseModel):
     """Response from deleting an evaluation"""
 
     eval_id: str
@@ -188,7 +189,7 @@ class DeleteEvalResponse(BaseModel):
     """Whether the evaluation was successfully deleted"""
 
 
-class CancelEvalResponse(BaseModel):
+class CancelEvalResponse(LiteLLMBaseModel):
     """Response from cancelling an evaluation"""
 
     id: str
@@ -287,7 +288,7 @@ class ListRunsParams(TypedDict, total=False):
 
 
 # Run API Response Types
-class ResultCounts(BaseModel):
+class ResultCounts(LiteLLMBaseModel):
     """Result counts for a run"""
 
     total: int
@@ -303,7 +304,7 @@ class ResultCounts(BaseModel):
     """Number of error results"""
 
 
-class PerTestingCriteriaResult(BaseModel):
+class PerTestingCriteriaResult(LiteLLMBaseModel):
     """Results for a specific testing criteria"""
 
     testing_criteria_index: int
@@ -316,7 +317,7 @@ class PerTestingCriteriaResult(BaseModel):
     """Average score for this criteria"""
 
 
-class Run(BaseModel):
+class Run(LiteLLMBaseModel):
     """Represents a run from the OpenAI Evals API"""
 
     id: str
@@ -371,7 +372,7 @@ class Run(BaseModel):
     """Error details if the run failed"""
 
 
-class ListRunsResponse(BaseModel):
+class ListRunsResponse(LiteLLMBaseModel):
     """Response from listing runs"""
 
     object: str = "list"
@@ -390,7 +391,7 @@ class ListRunsResponse(BaseModel):
     """Whether there are more runs available"""
 
 
-class CancelRunResponse(BaseModel):
+class CancelRunResponse(LiteLLMBaseModel):
     """Response from cancelling a run"""
 
     id: str
@@ -403,7 +404,7 @@ class CancelRunResponse(BaseModel):
     """Status after cancellation, always 'cancelled'"""
 
 
-class RunDeleteResponse(BaseModel):
+class RunDeleteResponse(LiteLLMBaseModel):
     """Response from deleting a run"""
 
     run_id: str
diff --git a/litellm/types/llms/vertex_ai_gemini_transcription.py b/litellm/types/llms/vertex_ai_gemini_transcription.py
index e039bc8f2eb..e78aafaca9f 100644
--- a/litellm/types/llms/vertex_ai_gemini_transcription.py
+++ b/litellm/types/llms/vertex_ai_gemini_transcription.py
@@ -1,8 +1,10 @@
 from typing import Literal
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 from typing_extensions import ReadOnly, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 
 class VertexGeminiTranscriptionInlineData(TypedDict):
     mimeType: ReadOnly[str]
@@ -31,32 +33,32 @@ class VertexGeminiTranscriptionRequest(TypedDict):
     generationConfig: ReadOnly[VertexGeminiTranscriptionGenerationConfig]
 
 
-class VertexGeminiTranscriptionResponsePart(BaseModel):
+class VertexGeminiTranscriptionResponsePart(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     text: str | None = None
 
 
-class VertexGeminiTranscriptionResponseContent(BaseModel):
+class VertexGeminiTranscriptionResponseContent(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     parts: tuple[VertexGeminiTranscriptionResponsePart, ...] = ()
 
 
-class VertexGeminiTranscriptionCandidate(BaseModel):
+class VertexGeminiTranscriptionCandidate(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     content: VertexGeminiTranscriptionResponseContent | None = None
 
 
-class VertexGeminiTranscriptionModalityTokens(BaseModel):
+class VertexGeminiTranscriptionModalityTokens(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     modality: str | None = None
     tokenCount: int = 0
 
 
-class VertexGeminiTranscriptionUsageMetadata(BaseModel):
+class VertexGeminiTranscriptionUsageMetadata(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     promptTokenCount: int = 0
@@ -65,7 +67,7 @@ class VertexGeminiTranscriptionUsageMetadata(BaseModel):
     promptTokensDetails: tuple[VertexGeminiTranscriptionModalityTokens, ...] = ()
 
 
-class VertexGeminiTranscriptionResponse(BaseModel):
+class VertexGeminiTranscriptionResponse(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     candidates: tuple[VertexGeminiTranscriptionCandidate, ...] = ()
diff --git a/litellm/types/llms/vertex_ai_gemma.py b/litellm/types/llms/vertex_ai_gemma.py
index f64f4d1d4fc..06fa64bbe4f 100644
--- a/litellm/types/llms/vertex_ai_gemma.py
+++ b/litellm/types/llms/vertex_ai_gemma.py
@@ -1,9 +1,11 @@
 from typing import Annotated, Literal
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class VertexGemmaContainerError(BaseModel):
+class VertexGemmaContainerError(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     object: Literal["error"]
     message: str
diff --git a/litellm/types/llms/vertex_ai_speech_to_text.py b/litellm/types/llms/vertex_ai_speech_to_text.py
index d07a5bbc192..f3d343d13ce 100644
--- a/litellm/types/llms/vertex_ai_speech_to_text.py
+++ b/litellm/types/llms/vertex_ai_speech_to_text.py
@@ -1,8 +1,10 @@
 from typing import Annotated, Literal
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
 from typing_extensions import TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 
 class VertexSpeechToTextAutoDecodingConfig(TypedDict):
     pass
@@ -24,25 +26,25 @@ class VertexSpeechToTextRecognizeRequest(TypedDict):
     content: str
 
 
-class VertexSpeechToTextAlternative(BaseModel):
+class VertexSpeechToTextAlternative(LiteLLMBaseModel):
     transcript: str | None = None
 
 
-class VertexSpeechToTextResult(BaseModel):
-    alternatives: list[VertexSpeechToTextAlternative] = []
+class VertexSpeechToTextResult(LiteLLMBaseModel):
+    alternatives: list[VertexSpeechToTextAlternative] = Field(default=[])
     languageCode: str | None = None
 
 
-class VertexSpeechToTextResponseMetadata(BaseModel):
+class VertexSpeechToTextResponseMetadata(LiteLLMBaseModel):
     totalBilledDuration: str | None = None
 
 
-class VertexSpeechToTextRecognizeResponse(BaseModel):
-    results: list[VertexSpeechToTextResult] = []
+class VertexSpeechToTextRecognizeResponse(LiteLLMBaseModel):
+    results: list[VertexSpeechToTextResult] = Field(default=[])
     metadata: VertexSpeechToTextResponseMetadata | None = None
 
 
-class VertexSpeechStreamingConfigure(BaseModel):
+class VertexSpeechStreamingConfigure(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     kind: Literal["configure"] = "configure"
     model: str
@@ -50,12 +52,12 @@ class VertexSpeechStreamingConfigure(BaseModel):
     sample_rate_hertz: int
 
 
-class VertexSpeechStreamingFinishTurn(BaseModel):
+class VertexSpeechStreamingFinishTurn(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     kind: Literal["finish_turn"] = "finish_turn"
 
 
-class VertexSpeechStreamingDiscardTurn(BaseModel):
+class VertexSpeechStreamingDiscardTurn(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     kind: Literal["discard_turn"] = "discard_turn"
 
@@ -66,13 +68,13 @@ VertexSpeechStreamingCommandUnion = (
 VertexSpeechStreamingCommand = Annotated[VertexSpeechStreamingCommandUnion, Field(discriminator="kind")]
 
 
-class VertexSpeechStreamingResult(BaseModel):
+class VertexSpeechStreamingResult(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     transcript: str
     is_final: bool
 
 
-class VertexSpeechStreamingResponse(BaseModel):
+class VertexSpeechStreamingResponse(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     kind: Literal["response"] = "response"
     speech_event: Literal["none", "begin", "end"]
@@ -80,17 +82,17 @@ class VertexSpeechStreamingResponse(BaseModel):
     billed_seconds: float
 
 
-class VertexSpeechStreamingConfigured(BaseModel):
+class VertexSpeechStreamingConfigured(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     kind: Literal["configured"] = "configured"
 
 
-class VertexSpeechStreamingTurnFinished(BaseModel):
+class VertexSpeechStreamingTurnFinished(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     kind: Literal["turn_finished"] = "turn_finished"
 
 
-class VertexSpeechStreamingTurnDiscarded(BaseModel):
+class VertexSpeechStreamingTurnDiscarded(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
     kind: Literal["turn_discarded"] = "turn_discarded"
     billed_seconds: float
diff --git a/litellm/types/management_endpoints/auto_router_endpoints.py b/litellm/types/management_endpoints/auto_router_endpoints.py
index ded971f6705..0738f7bc737 100644
--- a/litellm/types/management_endpoints/auto_router_endpoints.py
+++ b/litellm/types/management_endpoints/auto_router_endpoints.py
@@ -7,9 +7,10 @@ from datetime import datetime, timezone
 from types import MappingProxyType
 from typing import Final, Literal, TypeAlias
 
-from pydantic import BaseModel, Field, computed_field, field_validator, model_validator
+from pydantic import Field, computed_field, field_validator, model_validator
 
 from litellm.router_strategy.complexity_router.config import ComplexityRouterConfig
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import StandardLoggingRoutingDecision
 
 DEFAULT_ROUTING_TEST_ROUTER_NAME: Final[str] = "auto_router_routing_test"
@@ -28,7 +29,7 @@ class RequestComplexityRouterConfig(ComplexityRouterConfig):
     )
 
 
-class ComplexityRouterConfigValidationRequest(BaseModel):
+class ComplexityRouterConfigValidationRequest(LiteLLMBaseModel):
     """A complexity-router config to validate without saving, so a form can surface the
     backend's own verdict inline instead of a raw 400 at write time."""
 
@@ -39,18 +40,18 @@ class ComplexityRouterConfigValidationRequest(BaseModel):
     )
 
 
-class ComplexityRouterConfigValidationResponse(BaseModel):
+class ComplexityRouterConfigValidationResponse(LiteLLMBaseModel):
     valid: bool
     error: str | None = None
 
 
-class AutoRouterAvailabilityRequest(BaseModel):
+class AutoRouterAvailabilityRequest(LiteLLMBaseModel):
     team_id: str | None = None
     saved_model_id: str | None = None
     complexity_router_config: Mapping[str, object] | None = None
 
 
-class AutoRouterAllowance(BaseModel):
+class AutoRouterAllowance(LiteLLMBaseModel):
     key: str
     limit: int | None
     remaining: int | None
@@ -58,12 +59,12 @@ class AutoRouterAllowance(BaseModel):
     available: bool = True
 
 
-class AutoRouterAvailabilityResponse(BaseModel):
+class AutoRouterAvailabilityResponse(LiteLLMBaseModel):
     allowances: tuple[AutoRouterAllowance, ...]
     error: str | None = None
 
 
-class AutoRouterRoutingTestRequest(BaseModel):
+class AutoRouterRoutingTestRequest(LiteLLMBaseModel):
     """A single request to classify against a complexity-router config that need not be saved yet.
 
     Carries the same fields the serving path carries, so a dry run classifies what a real turn
@@ -158,7 +159,7 @@ class AutoRouterRoutingTestRequest(BaseModel):
         )
 
 
-class AutoRouterRoutingTestResponse(BaseModel):
+class AutoRouterRoutingTestResponse(LiteLLMBaseModel):
     """Where one prompt would have been routed, and why."""
 
     routed_model: str = Field(description="The model group the router picked")
@@ -170,7 +171,7 @@ class AutoRouterRoutingTestResponse(BaseModel):
     )
 
 
-class AutoRouterCacheBucket(BaseModel):
+class AutoRouterCacheBucket(LiteLLMBaseModel):
     """One prompt-caching bucket of turns, with how often those turns hit the cache."""
 
     turns: int = Field(description="Turns classified into this bucket")
@@ -178,7 +179,7 @@ class AutoRouterCacheBucket(BaseModel):
     hit_rate_pct: float = Field(description="hits over this bucket's turns, as a percentage")
 
 
-class AutoRouterCacheStats(BaseModel):
+class AutoRouterCacheStats(LiteLLMBaseModel):
     """Prompt-caching behaviour of auto-routed turns, bucketed by what the router did.
 
     Every in-order turn falls in exactly one bucket: the session stayed on the same model,
@@ -204,7 +205,7 @@ class AutoRouterCacheStats(BaseModel):
     ttl_1h_turns: int = Field(description="Turns whose cache write used the one-hour TTL")
 
 
-class AutoRouterBenchmarkTotals(BaseModel):
+class AutoRouterBenchmarkTotals(LiteLLMBaseModel):
     """Auto-routed traffic in the window. Turns, spend and savings count requests on the selected UTC days;
     the session averages and cache stats describe every session overlapping the window, whole."""
 
@@ -266,7 +267,7 @@ class AutoRouterBenchmarkGroup(AutoRouterBenchmarkTotals):
     )
 
 
-class AutoRouterSessionResponse(BaseModel):
+class AutoRouterSessionResponse(LiteLLMBaseModel):
     """One auto-routed session as its own key sees it: what the last turn ran on, and what the session cost
     against the router's savings baseline (the priciest model in its hardest tier)."""
 
@@ -299,7 +300,7 @@ class AutoRouterSessionResponse(BaseModel):
     )
 
 
-class AutoRouterBenchmarksResponse(BaseModel):
+class AutoRouterBenchmarksResponse(LiteLLMBaseModel):
     """Benchmarks for the auto-router dashboard, aggregated from the per-session and per-day rollups."""
 
     start_date: str = Field(description="Window start day, YYYY-MM-DD UTC, inclusive")
@@ -335,7 +336,7 @@ SHADOW_EVAL_TURN_VALVE: Final[int] = 10_000
 SHADOW_EVAL_MAX_ROUTERS: Final[int] = 4
 
 
-class StartShadowEvalRequest(BaseModel):
+class StartShadowEvalRequest(LiteLLMBaseModel):
     """Start duplicating one or more targets' traffic for blind comparison against an auto-router.
 
     A target is a virtual key, a team, or a user; each becomes its own leg with its own
@@ -522,7 +523,7 @@ class StartShadowEvalRequest(BaseModel):
         return self
 
 
-class ShadowEvalSlice(BaseModel):
+class ShadowEvalSlice(LiteLLMBaseModel):
     """Judge outcomes for one slice of a job's verdicts: a router tier, one of the
     models that served the real arm, or one scoped target (embedded on that target's
     own entry, so slices never need re-joining to a target by id)."""
@@ -566,7 +567,7 @@ class ShadowEvalSlice(BaseModel):
     )
 
 
-class ShadowEvalResult(BaseModel):
+class ShadowEvalResult(LiteLLMBaseModel):
     """Stratified results of a shadow-eval job's verdicts so far."""
 
     by_tier: tuple[ShadowEvalSlice, ...]
@@ -624,7 +625,7 @@ class ShadowEvalResult(BaseModel):
     )
 
 
-class ShadowEvalJobTargetResponse(BaseModel):
+class ShadowEvalJobTargetResponse(LiteLLMBaseModel):
     """One target a job shadows (a key, team, or user), with its own budget and stop state."""
 
     target_type: ShadowEvalTargetType = Field(description="What kind of entity this entry scopes")
@@ -690,7 +691,7 @@ class ShadowEvalJobTargetResponse(BaseModel):
     )
 
 
-class ShadowEvalJobResponse(BaseModel):
+class ShadowEvalJobResponse(LiteLLMBaseModel):
     """A shadow-eval job over one or more targets, each with its own budget and stop state;
     status is derived from stopped_by, the targets' stop and budget state, and ends_at,
     never stored, so no writer anywhere can produce an inconsistent one. Aggregate
diff --git a/litellm/types/management_endpoints/cache_settings_endpoints.py b/litellm/types/management_endpoints/cache_settings_endpoints.py
index cb05f3a50ac..f6394f52c99 100644
--- a/litellm/types/management_endpoints/cache_settings_endpoints.py
+++ b/litellm/types/management_endpoints/cache_settings_endpoints.py
@@ -4,10 +4,10 @@ Types and field definitions for cache settings management endpoints
 
 from typing import Any, Final
 
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class CacheSettingsField(BaseModel):
+class CacheSettingsField(LiteLLMBaseModel):
     field_name: str
     field_type: str
     field_value: Any
diff --git a/litellm/types/management_endpoints/coordination_redis_endpoints.py b/litellm/types/management_endpoints/coordination_redis_endpoints.py
index d70a921a22f..5d429b670dc 100644
--- a/litellm/types/management_endpoints/coordination_redis_endpoints.py
+++ b/litellm/types/management_endpoints/coordination_redis_endpoints.py
@@ -4,14 +4,14 @@ Types and field definitions for coordination Redis settings management endpoints
 
 from typing import Final, Literal
 
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 CoordinationRedisSection = Literal["connection", "cluster", "sentinel"]
 
 CoordinationRedisSource = Literal["coordination_redis", "cache_backend", "environment"]
 
 
-class CoordinationRedisSettingsField(BaseModel):
+class CoordinationRedisSettingsField(LiteLLMBaseModel):
     field_name: str
     field_type: str
     field_value: object | None = None
diff --git a/litellm/types/management_endpoints/prompt_cache_prediction.py b/litellm/types/management_endpoints/prompt_cache_prediction.py
index 3789607b021..b337d53f1ef 100644
--- a/litellm/types/management_endpoints/prompt_cache_prediction.py
+++ b/litellm/types/management_endpoints/prompt_cache_prediction.py
@@ -1,12 +1,14 @@
 from collections.abc import Mapping
 from typing import Annotated, Literal, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, Field, JsonValue, StrictInt
+from pydantic import ConfigDict, Field, JsonValue, StrictInt
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 TokenCount: TypeAlias = Annotated[StrictInt, Field(ge=0)]
 
 
-class CacheTokenBuckets(BaseModel):
+class CacheTokenBuckets(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="forbid")
 
     uncached_input_tokens: TokenCount = 0
@@ -24,7 +26,7 @@ class CacheTokenBuckets(BaseModel):
         )
 
 
-class CacheEvidence(BaseModel):
+class CacheEvidence(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     observed_at: float
@@ -33,12 +35,12 @@ class CacheEvidence(BaseModel):
     confidence: Literal["observed"] = "observed"
 
 
-class CacheCostScenario(BaseModel):
+class CacheCostScenario(LiteLLMBaseModel):
     tokens: CacheTokenBuckets
     input_cost: float
 
 
-class CachePredictionArm(BaseModel):
+class CachePredictionArm(LiteLLMBaseModel):
     deployment_id: str
     model: str | None = None
     cache_state: Literal["warm", "partial", "stale", "unknown", "disabled"] = "unknown"
@@ -50,7 +52,7 @@ class CachePredictionArm(BaseModel):
     token_count_source: Literal["anthropic_count_tokens"] | None = None
 
 
-class CachePredictionRequest(BaseModel):
+class CachePredictionRequest(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid")
 
     current_deployment_id: str = Field(min_length=1, max_length=256)
@@ -58,7 +60,7 @@ class CachePredictionRequest(BaseModel):
     request: Mapping[str, JsonValue]
 
 
-class CachePredictionResponse(BaseModel):
+class CachePredictionResponse(LiteLLMBaseModel):
     stay: CachePredictionArm
     switch: CachePredictionArm
     switch_delta: float | None
diff --git a/litellm/types/management_endpoints/prompt_caching_requests.py b/litellm/types/management_endpoints/prompt_caching_requests.py
index e72183a113b..fbf5f20826e 100644
--- a/litellm/types/management_endpoints/prompt_caching_requests.py
+++ b/litellm/types/management_endpoints/prompt_caching_requests.py
@@ -1,12 +1,14 @@
 from datetime import datetime
 from typing import Literal, TypeAlias
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 PromptCachingRequestFilter: TypeAlias = Literal["all", "injected", "hits"]
 
 
-class PromptCachingRequest(BaseModel):
+class PromptCachingRequest(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     request_id: str
@@ -19,14 +21,14 @@ class PromptCachingRequest(BaseModel):
     net_savings: float | None
 
 
-class PromptCachingRequestCursor(BaseModel):
+class PromptCachingRequestCursor(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     start_time: datetime
     request_id: str
 
 
-class PromptCachingRequestsResponse(BaseModel):
+class PromptCachingRequestsResponse(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     requests: tuple[PromptCachingRequest, ...]
diff --git a/litellm/types/management_endpoints/router_settings_endpoints.py b/litellm/types/management_endpoints/router_settings_endpoints.py
index 4d1ac58edb0..9c8fe446d72 100644
--- a/litellm/types/management_endpoints/router_settings_endpoints.py
+++ b/litellm/types/management_endpoints/router_settings_endpoints.py
@@ -4,12 +4,14 @@ Types and field definitions for router settings management endpoints
 
 from typing import Any, Final, Literal
 
-from pydantic import BaseModel, Field, field_validator
+from pydantic import Field, field_validator
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 # Fallback Management Types
 
 
-class FallbackCreateRequest(BaseModel):
+class FallbackCreateRequest(LiteLLMBaseModel):
     """Request model for creating/updating fallbacks"""
 
     model: str = Field(description="The model name to configure fallbacks for (e.g., 'gpt-3.5-turbo')")
@@ -39,7 +41,7 @@ class FallbackCreateRequest(BaseModel):
         return v.strip()
 
 
-class FallbackResponse(BaseModel):
+class FallbackResponse(LiteLLMBaseModel):
     """Response model for fallback operations"""
 
     model: str = Field(description="The model name")
@@ -48,7 +50,7 @@ class FallbackResponse(BaseModel):
     message: str = Field(description="Success message")
 
 
-class FallbackGetResponse(BaseModel):
+class FallbackGetResponse(LiteLLMBaseModel):
     """Response model for getting fallbacks"""
 
     model: str = Field(description="The model name")
@@ -56,7 +58,7 @@ class FallbackGetResponse(BaseModel):
     fallback_type: str = Field(description="Type of fallback")
 
 
-class FallbackDeleteResponse(BaseModel):
+class FallbackDeleteResponse(LiteLLMBaseModel):
     """Response model for deleting fallbacks"""
 
     model: str = Field(description="The model name")
@@ -67,7 +69,7 @@ class FallbackDeleteResponse(BaseModel):
 # Router Settings Types
 
 
-class RouterSettingsField(BaseModel):
+class RouterSettingsField(LiteLLMBaseModel):
     field_name: str
     field_type: str
     field_value: Any
diff --git a/litellm/types/mcp.py b/litellm/types/mcp.py
index 949506898b1..556cb6712ae 100644
--- a/litellm/types/mcp.py
+++ b/litellm/types/mcp.py
@@ -8,10 +8,10 @@ from typing import TYPE_CHECKING, Annotated, Any, Final, Literal
 from urllib.parse import urlsplit
 
 import httpx
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
 from typing_extensions import ReadOnly, TypedDict
 
-from litellm.types.llms.base import HiddenParams
+from litellm.types.llms.base import HiddenParams, LiteLLMBaseModel
 
 if TYPE_CHECKING:
     import httpx2
@@ -98,7 +98,7 @@ MCPAuthType = (
 )
 
 
-class MCPPublicServer(BaseModel):
+class MCPPublicServer(LiteLLMBaseModel):
     """
     Safe params for public MCP servers
     """
@@ -113,7 +113,7 @@ class MCPPublicServer(BaseModel):
     mcp_info: dict[str, Any] | None = None
 
 
-class MCPAllowedClient(BaseModel):
+class MCPAllowedClient(LiteLLMBaseModel):
     """One entry of `general_settings.mcp_allowed_clients`."""
 
     model_config = ConfigDict(frozen=True)
@@ -129,7 +129,7 @@ class MCPAllowedClient(BaseModel):
     )
 
 
-class MCPToolSearchSettings(BaseModel):
+class MCPToolSearchSettings(LiteLLMBaseModel):
     """`litellm_settings.mcp_tool_search`: how the native `mcp_tool_search` virtual tool ranks the caller's tools."""
 
     model_config = ConfigDict(frozen=True)
@@ -424,7 +424,7 @@ class MCPStdioConfig(TypedDict, total=False):
     """
 
 
-class MCPPreCallRequestObject(BaseModel):
+class MCPPreCallRequestObject(LiteLLMBaseModel):
     """
     Pydantic object used for MCP pre_call_hook request validation and modification
     """
@@ -438,7 +438,7 @@ class MCPPreCallRequestObject(BaseModel):
     hidden_params: HiddenParams = HiddenParams()
 
 
-class MCPPreCallResponseObject(BaseModel):
+class MCPPreCallResponseObject(LiteLLMBaseModel):
     """
     Pydantic object used for MCP pre_call_hook response
     """
@@ -449,7 +449,7 @@ class MCPPreCallResponseObject(BaseModel):
     hidden_params: HiddenParams = HiddenParams()
 
 
-class MCPDuringCallRequestObject(BaseModel):
+class MCPDuringCallRequestObject(LiteLLMBaseModel):
     """
     Pydantic object used for MCP during_call_hook request
     """
@@ -463,7 +463,7 @@ class MCPDuringCallRequestObject(BaseModel):
     hidden_params: HiddenParams = HiddenParams()
 
 
-class MCPDuringCallResponseObject(BaseModel):
+class MCPDuringCallResponseObject(LiteLLMBaseModel):
     """
     Pydantic object used for MCP during_call_hook response
     """
@@ -473,7 +473,7 @@ class MCPDuringCallResponseObject(BaseModel):
     hidden_params: HiddenParams = HiddenParams()
 
 
-class MCPPostCallResponseObject(BaseModel):
+class MCPPostCallResponseObject(LiteLLMBaseModel):
     """
     Pydantic object used for MCP post_call_hook response
     """
@@ -482,7 +482,7 @@ class MCPPostCallResponseObject(BaseModel):
     hidden_params: HiddenParams
 
 
-class MCPGatewaySession(BaseModel):
+class MCPGatewaySession(LiteLLMBaseModel):
     """One live stateful Streamable HTTP session held by this proxy worker."""
 
     session_id_prefix: str
@@ -498,12 +498,12 @@ class MCPGatewaySession(BaseModel):
     in_flight_requests: int
 
 
-class MCPGatewaySessionGroupCount(BaseModel):
+class MCPGatewaySessionGroupCount(LiteLLMBaseModel):
     label: str | None = None
     count: int
 
 
-class MCPGatewaySessionsResponse(BaseModel):
+class MCPGatewaySessionsResponse(LiteLLMBaseModel):
     worker_pid: int
     total_sessions: int
     by_client: list[MCPGatewaySessionGroupCount] = Field(default_factory=list)
@@ -511,7 +511,7 @@ class MCPGatewaySessionsResponse(BaseModel):
     sessions: list[MCPGatewaySession] = Field(default_factory=list)
 
 
-class MCPGatewaySessionsTerminateResponse(BaseModel):
+class MCPGatewaySessionsTerminateResponse(LiteLLMBaseModel):
     """Stateful sessions an administrator force-closed on this proxy worker."""
 
     worker_pid: int
diff --git a/litellm/types/mcp_server/mcp_server_manager.py b/litellm/types/mcp_server/mcp_server_manager.py
index 343fb731b06..c54eda505df 100644
--- a/litellm/types/mcp_server/mcp_server_manager.py
+++ b/litellm/types/mcp_server/mcp_server_manager.py
@@ -2,9 +2,10 @@ import json
 from datetime import datetime
 from typing import Annotated, Any, Final, Literal
 
-from pydantic import AfterValidator, BaseModel, ConfigDict, Field, TypeAdapter, field_validator, model_validator
+from pydantic import AfterValidator, ConfigDict, Field, TypeAdapter, field_validator, model_validator
 from typing_extensions import Self
 
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.mcp import (
     DEFAULT_SUBJECT_TOKEN_TYPE,
     MCPAuth,
@@ -27,7 +28,7 @@ def _validate_mcp_protocol_metadata(value: dict[str, object]) -> dict[str, objec
 MCPInfo = Annotated[dict[str, Any], AfterValidator(_validate_mcp_protocol_metadata)]
 
 
-class MCPOAuthMetadata(BaseModel):
+class MCPOAuthMetadata(LiteLLMBaseModel):
     scopes: list[str] | None = None
     """Resource-driven scopes for the authorization request: the RFC 9728 protected-resource
     ``scopes_supported``, or the ``scope`` from the WWW-Authenticate 401 challenge when the resource
@@ -50,7 +51,7 @@ class MCPOAuthMetadata(BaseModel):
     usable in memory but must never be persisted as configuration."""
 
 
-class MCPOAuthIdentityBinding(BaseModel):
+class MCPOAuthIdentityBinding(LiteLLMBaseModel):
     """Per-server policy binding stored per-user OAuth credentials to the authenticated LiteLLM caller.
 
     When enabled for an interactive oauth2 server, the token relay validates the upstream OIDC
@@ -70,7 +71,7 @@ class MCPOAuthIdentityBinding(BaseModel):
     require_email_verified: bool = True
 
 
-class PinnedMCPTool(BaseModel):
+class PinnedMCPTool(LiteLLMBaseModel):
     """One tool of an admin-pinned catalog: the description and input schema tools/list keeps serving."""
 
     model_config = ConfigDict(frozen=True, extra="forbid")
@@ -87,7 +88,7 @@ def parse_pinned_tools(value: object) -> dict[str, PinnedMCPTool] | None:
     return _PINNED_TOOLS.validate_python(decoded or None)
 
 
-class MCPServer(BaseModel):
+class MCPServer(LiteLLMBaseModel):
     server_id: str
     name: str
     alias: str | None = None
@@ -213,7 +214,7 @@ class MCPServer(BaseModel):
     dcr_bridge: bool | None = None
     per_server_oauth_discovery: bool = False
     is_byok: bool = False
-    byok_description: list[str] = []
+    byok_description: list[str] = Field(default=[])
     byok_api_key_help_url: str | None = None
     source_url: str | None = None
     created_at: datetime | None = None
diff --git a/litellm/types/mcp_server/mcp_toolset.py b/litellm/types/mcp_server/mcp_toolset.py
index 7e9e03e48ed..2eab0af161f 100644
--- a/litellm/types/mcp_server/mcp_toolset.py
+++ b/litellm/types/mcp_server/mcp_toolset.py
@@ -1,32 +1,34 @@
 from datetime import datetime
 
-from pydantic import BaseModel
+from pydantic import Field
 from typing_extensions import TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 
 class MCPToolsetTool(TypedDict):
     server_id: str
     tool_name: str
 
 
-class MCPToolset(BaseModel):
+class MCPToolset(LiteLLMBaseModel):
     toolset_id: str
     toolset_name: str
     description: str | None = None
-    tools: list[MCPToolsetTool] = []
+    tools: list[MCPToolsetTool] = Field(default=[])
     created_at: datetime | None = None
     created_by: str | None = None
     updated_at: datetime | None = None
     updated_by: str | None = None
 
 
-class NewMCPToolsetRequest(BaseModel):
+class NewMCPToolsetRequest(LiteLLMBaseModel):
     toolset_name: str
     description: str | None = None
-    tools: list[MCPToolsetTool] = []
+    tools: list[MCPToolsetTool] = Field(default=[])
 
 
-class UpdateMCPToolsetRequest(BaseModel):
+class UpdateMCPToolsetRequest(LiteLLMBaseModel):
     toolset_id: str
     toolset_name: str | None = None
     description: str | None = None
diff --git a/litellm/types/mcp_server/tool_registry.py b/litellm/types/mcp_server/tool_registry.py
index 79dedfd7653..df2dda27a38 100644
--- a/litellm/types/mcp_server/tool_registry.py
+++ b/litellm/types/mcp_server/tool_registry.py
@@ -1,10 +1,12 @@
 from collections.abc import Callable
 from typing import Any, ClassVar
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class MCPTool(BaseModel):
+class MCPTool(LiteLLMBaseModel):
     model_config: ClassVar[ConfigDict] = ConfigDict(arbitrary_types_allowed=True)
     name: str
     description: str
@@ -12,23 +14,23 @@ class MCPTool(BaseModel):
     handler: Callable
 
 
-class ToolSchema(BaseModel):
+class ToolSchema(LiteLLMBaseModel):
     name: str
     description: str
     inputSchema: dict[str, Any]
 
 
-class ListToolsResponse(BaseModel):
+class ListToolsResponse(LiteLLMBaseModel):
     tools: list[ToolSchema]
     nextCursor: str | None = None
     _meta: dict[str, Any] | None = None
 
 
-class CallToolRequest(BaseModel):
+class CallToolRequest(LiteLLMBaseModel):
     method: str = "tools/call"
     params: dict[str, Any]
 
 
-class ContentItem(BaseModel):
+class ContentItem(LiteLLMBaseModel):
     type: str
     text: str | None = None
diff --git a/litellm/types/memory_management.py b/litellm/types/memory_management.py
index 153de0c6cb9..ad2b328fd1c 100644
--- a/litellm/types/memory_management.py
+++ b/litellm/types/memory_management.py
@@ -4,10 +4,12 @@ Pydantic models for Memory management endpoints.
 
 from datetime import datetime
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class LiteLLM_MemoryRow(BaseModel):
+class LiteLLM_MemoryRow(LiteLLMBaseModel):
     memory_id: str
     key: str
     value: str
@@ -20,7 +22,7 @@ class LiteLLM_MemoryRow(BaseModel):
     updated_by: str | None = None
 
 
-class MemoryCreateRequest(BaseModel):
+class MemoryCreateRequest(LiteLLMBaseModel):
     key: str = Field(..., description="Memory key (acts as the namespace in the URL).")
     value: str = Field(..., description="Memory content. Typically markdown/text for LLM context.")
     metadata: object | None = Field(
@@ -37,7 +39,7 @@ class MemoryCreateRequest(BaseModel):
     )
 
 
-class MemoryUpdateRequest(BaseModel):
+class MemoryUpdateRequest(LiteLLMBaseModel):
     value: str | None = None
     metadata: object | None = None
     # Only honored on create (when the row doesn't yet exist) and only for
@@ -47,11 +49,11 @@ class MemoryUpdateRequest(BaseModel):
     team_id: str | None = None
 
 
-class MemoryListResponse(BaseModel):
+class MemoryListResponse(LiteLLMBaseModel):
     memories: list[LiteLLM_MemoryRow]
     total: int
 
 
-class MemoryDeleteResponse(BaseModel):
+class MemoryDeleteResponse(LiteLLMBaseModel):
     key: str
     deleted: bool
diff --git a/litellm/types/model_insights.py b/litellm/types/model_insights.py
index 8d4fbbff9d3..5b5a9efed37 100644
--- a/litellm/types/model_insights.py
+++ b/litellm/types/model_insights.py
@@ -1,11 +1,11 @@
 from typing import Literal
 
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 ModelInsightsMetric = Literal["requests", "spend", "tokens"]
 
 
-class ModelInsightMetric(BaseModel):
+class ModelInsightMetric(LiteLLMBaseModel):
     model_group: str
     model: str
     provider: str
@@ -21,7 +21,7 @@ class ModelInsightDailyMetric(ModelInsightMetric):
     date: str
 
 
-class ModelInsightDailyTotal(BaseModel):
+class ModelInsightDailyTotal(LiteLLMBaseModel):
     date: str
     spend: float
     prompt_tokens: int
@@ -29,7 +29,7 @@ class ModelInsightDailyTotal(BaseModel):
     requests: int
 
 
-class ModelInsightTask(BaseModel):
+class ModelInsightTask(LiteLLMBaseModel):
     task_type: str
     label: str
     category: str
@@ -42,7 +42,7 @@ class ModelInsightTaskSummary(ModelInsightTask):
     provider: str
 
 
-class ModelInsightsResponse(BaseModel):
+class ModelInsightsResponse(LiteLLMBaseModel):
     start_date: str
     end_date: str
     daily: list[ModelInsightDailyMetric]
@@ -50,7 +50,7 @@ class ModelInsightsResponse(BaseModel):
     top_models: list[ModelInsightMetric]
 
 
-class ModelInsightTasksResponse(BaseModel):
+class ModelInsightTasksResponse(LiteLLMBaseModel):
     start_date: str
     end_date: str
     tasks: list[ModelInsightTaskSummary]
diff --git a/litellm/types/passthrough_endpoints/vertex_ai.py b/litellm/types/passthrough_endpoints/vertex_ai.py
index d1affd7be2c..e350470881f 100644
--- a/litellm/types/passthrough_endpoints/vertex_ai.py
+++ b/litellm/types/passthrough_endpoints/vertex_ai.py
@@ -2,12 +2,12 @@
 Used for /vertex_ai/ pass through endpoints
 """
 
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from ..llms.vertex_ai import VERTEX_CREDENTIALS_TYPES
 
 
-class VertexPassThroughCredentials(BaseModel):
+class VertexPassThroughCredentials(LiteLLMBaseModel):
     # Example: vertex_project = "my-project-123"
     vertex_project: str | None = None
 
diff --git a/litellm/types/prompts/init_prompts.py b/litellm/types/prompts/init_prompts.py
index f1a926f9ae0..fc4c36e760a 100644
--- a/litellm/types/prompts/init_prompts.py
+++ b/litellm/types/prompts/init_prompts.py
@@ -2,7 +2,9 @@ from datetime import datetime
 from enum import Enum
 from typing import Any, Literal
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
 class SupportedPromptIntegrations(str, Enum):
@@ -15,14 +17,14 @@ class SupportedPromptIntegrations(str, Enum):
     ARIZE_PHOENIX = "arize_phoenix"
 
 
-class PromptInfo(BaseModel):
+class PromptInfo(LiteLLMBaseModel):
     prompt_type: Literal["config", "db"]
     environment: str | None = "development"
 
     model_config = ConfigDict(extra="allow", protected_namespaces=())
 
 
-class PromptLiteLLMParams(BaseModel):
+class PromptLiteLLMParams(LiteLLMBaseModel):
     prompt_id: str | None = None
     prompt_integration: str
 
@@ -42,7 +44,7 @@ class PromptLiteLLMParams(BaseModel):
     model_config = ConfigDict(extra="allow", protected_namespaces=())
 
 
-class PromptSpec(BaseModel):
+class PromptSpec(LiteLLMBaseModel):
     prompt_id: str
     litellm_params: PromptLiteLLMParams
     prompt_info: PromptInfo
@@ -61,17 +63,17 @@ class PromptSpec(BaseModel):
         super().__init__(**data)
 
 
-class PromptTemplateBase(BaseModel):
+class PromptTemplateBase(LiteLLMBaseModel):
     litellm_prompt_id: str
     content: str
     metadata: dict[str, Any] | None = None
 
 
-class PromptInfoResponse(BaseModel):
+class PromptInfoResponse(LiteLLMBaseModel):
     prompt_spec: PromptSpec
     raw_prompt_template: PromptTemplateBase | None = None
     environments: list[str] | None = None  # All environments this prompt is deployed to
 
 
-class ListPromptsResponse(BaseModel):
+class ListPromptsResponse(LiteLLMBaseModel):
     prompts: list[PromptSpec]
diff --git a/litellm/types/proxy/agent_identity.py b/litellm/types/proxy/agent_identity.py
index a7fe0be37e1..4492773cdc0 100644
--- a/litellm/types/proxy/agent_identity.py
+++ b/litellm/types/proxy/agent_identity.py
@@ -2,12 +2,14 @@ from datetime import datetime
 from typing import Literal, TypeAlias
 from uuid import UUID
 
-from pydantic import BaseModel, ConfigDict, Field, field_validator
+from pydantic import ConfigDict, Field, field_validator
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 AgentExecutionMode: TypeAlias = Literal["autonomous", "delegated", "both"]
 
 
-class EntraIdentityConfig(BaseModel):
+class EntraIdentityConfig(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="forbid")
 
     provider: Literal["microsoft_entra"]
@@ -30,7 +32,7 @@ class EntraIdentityConfig(BaseModel):
         return f"https://login.microsoftonline.com/{self.tenant_id}/v2.0"
 
 
-class AgentIdentityBinding(BaseModel):
+class AgentIdentityBinding(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     agent_id: str
@@ -46,7 +48,7 @@ class AgentIdentityBinding(BaseModel):
     last_authenticated_at: datetime | None = None
 
 
-class AgentSubject(BaseModel):
+class AgentSubject(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     kind: Literal["application", "delegated_subject"]
@@ -54,14 +56,14 @@ class AgentSubject(BaseModel):
     mode: Literal["autonomous", "delegated"]
 
 
-class AgentIdentityFailure(BaseModel):
+class AgentIdentityFailure(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     code: Literal["identity_denied", "policy_unavailable"] = "identity_denied"
     message: str
 
 
-class ManagedAgentContext(BaseModel):
+class ManagedAgentContext(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     agent_id: str
@@ -71,7 +73,7 @@ class ManagedAgentContext(BaseModel):
     subject_oid: str | None = None
 
 
-class VerifiedHumanSubject(BaseModel):
+class VerifiedHumanSubject(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     issuer: str
@@ -80,7 +82,7 @@ class VerifiedHumanSubject(BaseModel):
     user_id: str
 
 
-class MicrosoftInteractiveSubject(BaseModel):
+class MicrosoftInteractiveSubject(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     issuer: str
@@ -88,7 +90,7 @@ class MicrosoftInteractiveSubject(BaseModel):
     oid: str
 
 
-class ManagedAgentIdentityStatus(BaseModel):
+class ManagedAgentIdentityStatus(LiteLLMBaseModel):
     identity: AgentIdentityBinding | None = None
     identity_managed: bool = False
     enabled: bool = True
diff --git a/litellm/types/proxy/callback_logs_endpoints.py b/litellm/types/proxy/callback_logs_endpoints.py
index 0ff1d5a2273..b2519d97122 100644
--- a/litellm/types/proxy/callback_logs_endpoints.py
+++ b/litellm/types/proxy/callback_logs_endpoints.py
@@ -7,12 +7,13 @@ payloads here; the proxy replays them through the standard callback fan-out.
 
 from typing import Any, Literal
 
-from pydantic import BaseModel, Field
+from pydantic import Field
 
 from litellm.constants import MAX_CALLBACK_LOG_RECORDS
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class CallbackLogRecord(BaseModel):
+class CallbackLogRecord(LiteLLMBaseModel):
     """A single finished logging event to replay through the callbacks."""
 
     status: Literal["success", "failure"]
@@ -20,7 +21,7 @@ class CallbackLogRecord(BaseModel):
     error: str | None = None
 
 
-class CallbackLogsRequest(BaseModel):
+class CallbackLogsRequest(LiteLLMBaseModel):
     """A batch of logging events posted by an external producer."""
 
     # Bounded so one POST can't trigger an unbounded callback/DB fan-out (each
@@ -28,14 +29,14 @@ class CallbackLogsRequest(BaseModel):
     records: list[CallbackLogRecord] = Field(..., max_length=MAX_CALLBACK_LOG_RECORDS)
 
 
-class CallbackLogFailure(BaseModel):
+class CallbackLogFailure(LiteLLMBaseModel):
     """A record that failed to replay, identified by its index in the batch."""
 
     index: int
     error: str
 
 
-class CallbackLogsResponse(BaseModel):
+class CallbackLogsResponse(LiteLLMBaseModel):
     """Per-batch result: counts plus per-record failure detail so the caller can
     distinguish a transient callback error from a structurally bad payload."""
 
diff --git a/litellm/types/proxy/carried_budget_state.py b/litellm/types/proxy/carried_budget_state.py
index 0e64b91645d..1a0c55a38ff 100644
--- a/litellm/types/proxy/carried_budget_state.py
+++ b/litellm/types/proxy/carried_budget_state.py
@@ -10,11 +10,13 @@ from collections.abc import Mapping
 from datetime import datetime
 from types import MappingProxyType
 
-from pydantic import BaseModel, ConfigDict, Field, ValidationError
+from pydantic import ConfigDict, Field, ValidationError
 from typing_extensions import Self
 
+from litellm.types.llms.base import LiteLLMBaseModel
 
-class _BudgetSnapshot(BaseModel):
+
+class _BudgetSnapshot(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, populate_by_name=True, extra="ignore")
 
     def metadata_entries(self) -> Mapping[str, object]:
diff --git a/litellm/types/proxy/claude_code_endpoints.py b/litellm/types/proxy/claude_code_endpoints.py
index dcb5561cfeb..28285838637 100644
--- a/litellm/types/proxy/claude_code_endpoints.py
+++ b/litellm/types/proxy/claude_code_endpoints.py
@@ -2,24 +2,26 @@
 Claude Code Marketplace endpoint types for LiteLLM Proxy
 """
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class PluginAuthor(BaseModel):
+class PluginAuthor(LiteLLMBaseModel):
     """Plugin author information."""
 
     name: str = Field(..., description="Author name")
     email: str | None = Field(None, description="Author email")
 
 
-class PluginOwner(BaseModel):
+class PluginOwner(LiteLLMBaseModel):
     """Marketplace owner information."""
 
     name: str = Field(..., description="Owner name")
     email: str | None = Field(None, description="Owner email")
 
 
-class PluginSpec(BaseModel):
+class PluginSpec(LiteLLMBaseModel):
     """Mutable fields shared by plugin create and update requests."""
 
     source: dict[str, str] = Field(
@@ -71,7 +73,7 @@ class UpdatePluginRequest(PluginSpec):
     version: str | None = Field(None, description="Semantic version; cleared if omitted")
 
 
-class PluginResponse(BaseModel):
+class PluginResponse(LiteLLMBaseModel):
     """Plugin information in API responses."""
 
     id: str = Field(..., description="Plugin unique ID")
@@ -82,7 +84,7 @@ class PluginResponse(BaseModel):
     enabled: bool = Field(..., description="Whether plugin is enabled")
 
 
-class RegisterPluginResponse(BaseModel):
+class RegisterPluginResponse(LiteLLMBaseModel):
     """Response from plugin registration."""
 
     status: str = Field(..., description="Operation status")
@@ -90,7 +92,7 @@ class RegisterPluginResponse(BaseModel):
     plugin: PluginResponse = Field(..., description="Plugin information")
 
 
-class PluginListItem(BaseModel):
+class PluginListItem(LiteLLMBaseModel):
     """Plugin item in list responses."""
 
     id: str
@@ -109,14 +111,14 @@ class PluginListItem(BaseModel):
     updated_at: str | None
 
 
-class ListPluginsResponse(BaseModel):
+class ListPluginsResponse(LiteLLMBaseModel):
     """Response from listing plugins."""
 
     plugins: list[PluginListItem]
     count: int
 
 
-class MarketplacePluginEntry(BaseModel):
+class MarketplacePluginEntry(LiteLLMBaseModel):
     """Plugin entry in marketplace.json."""
 
     name: str
@@ -129,7 +131,7 @@ class MarketplacePluginEntry(BaseModel):
     category: str | None = None
 
 
-class MarketplaceResponse(BaseModel):
+class MarketplaceResponse(LiteLLMBaseModel):
     """
     Marketplace catalog response.
 
diff --git a/litellm/types/proxy/cloudzero_endpoints.py b/litellm/types/proxy/cloudzero_endpoints.py
index 70b779338f8..ece294674b9 100644
--- a/litellm/types/proxy/cloudzero_endpoints.py
+++ b/litellm/types/proxy/cloudzero_endpoints.py
@@ -5,10 +5,12 @@ CloudZero endpoint types for LiteLLM Proxy
 from datetime import datetime
 from typing import Any
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class CloudZeroInitRequest(BaseModel):
+class CloudZeroInitRequest(LiteLLMBaseModel):
     """Request model for initializing CloudZero settings"""
 
     api_key: str = Field(..., description="CloudZero API key for authentication")
@@ -16,14 +18,14 @@ class CloudZeroInitRequest(BaseModel):
     timezone: str = Field(default="UTC", description="Timezone for date handling (default: UTC)")
 
 
-class CloudZeroInitResponse(BaseModel):
+class CloudZeroInitResponse(LiteLLMBaseModel):
     """Response model for CloudZero initialization"""
 
     message: str
     status: str
 
 
-class CloudZeroExportRequest(BaseModel):
+class CloudZeroExportRequest(LiteLLMBaseModel):
     """Request model for CloudZero export operations"""
 
     limit: int | None = Field(None, description="Optional limit on number of records to export")
@@ -35,7 +37,7 @@ class CloudZeroExportRequest(BaseModel):
     end_time_utc: datetime | None = Field(None, description="End time for data export in UTC")
 
 
-class CloudZeroExportResponse(BaseModel):
+class CloudZeroExportResponse(LiteLLMBaseModel):
     """Response model for CloudZero export operations"""
 
     message: str
@@ -47,7 +49,7 @@ class CloudZeroExportResponse(BaseModel):
     summary: dict[str, Any] | None = Field(None, description="Summary statistics for dry run")
 
 
-class CloudZeroSettingsView(BaseModel):
+class CloudZeroSettingsView(LiteLLMBaseModel):
     """Response model for viewing CloudZero settings with masked API key"""
 
     api_key_masked: str | None = Field(None, description="Masked API key showing only first 4 and last 4 characters")
@@ -56,7 +58,7 @@ class CloudZeroSettingsView(BaseModel):
     status: str | None = Field(None, description="Configuration status")
 
 
-class CloudZeroSettingsUpdate(BaseModel):
+class CloudZeroSettingsUpdate(LiteLLMBaseModel):
     """Request model for updating CloudZero settings"""
 
     api_key: str | None = Field(None, description="New CloudZero API key for authentication")
diff --git a/litellm/types/proxy/compliance_endpoints.py b/litellm/types/proxy/compliance_endpoints.py
index 0c1a11bf594..ef3fc73ef9a 100644
--- a/litellm/types/proxy/compliance_endpoints.py
+++ b/litellm/types/proxy/compliance_endpoints.py
@@ -1,7 +1,7 @@
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class ComplianceCheckResult(BaseModel):
+class ComplianceCheckResult(LiteLLMBaseModel):
     """Result of a single compliance check."""
 
     check_name: str
@@ -10,7 +10,7 @@ class ComplianceCheckResult(BaseModel):
     detail: str
 
 
-class ComplianceResponse(BaseModel):
+class ComplianceResponse(LiteLLMBaseModel):
     """Response from a compliance check endpoint."""
 
     compliant: bool
@@ -18,7 +18,7 @@ class ComplianceResponse(BaseModel):
     checks: list[ComplianceCheckResult]
 
 
-class ComplianceCheckRequest(BaseModel):
+class ComplianceCheckRequest(LiteLLMBaseModel):
     """Request payload for compliance check endpoints.
 
     Mirrors the spend log fields needed for compliance evaluation.
diff --git a/litellm/types/proxy/control_plane_endpoints.py b/litellm/types/proxy/control_plane_endpoints.py
index 8bf4c44b205..d75fd0a38b4 100644
--- a/litellm/types/proxy/control_plane_endpoints.py
+++ b/litellm/types/proxy/control_plane_endpoints.py
@@ -1,7 +1,9 @@
-from pydantic import BaseModel, field_validator
+from pydantic import field_validator
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class WorkerRegistryEntry(BaseModel):
+class WorkerRegistryEntry(LiteLLMBaseModel):
     worker_id: str
     name: str
     url: str
diff --git a/litellm/types/proxy/discovery_endpoints/agent_skills_endpoints.py b/litellm/types/proxy/discovery_endpoints/agent_skills_endpoints.py
index 0d8bb29e172..c3275c7a1cc 100644
--- a/litellm/types/proxy/discovery_endpoints/agent_skills_endpoints.py
+++ b/litellm/types/proxy/discovery_endpoints/agent_skills_endpoints.py
@@ -5,14 +5,16 @@ Schema: https://schemas.agentskills.io/discovery/0.2.0/schema.json
 
 from typing import Final, Literal
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 AGENT_SKILLS_DISCOVERY_SCHEMA_URL: Final = "https://schemas.agentskills.io/discovery/0.2.0/schema.json"
 MAX_SKILL_NAME_LENGTH: Final = 64
 MAX_SKILL_DESCRIPTION_LENGTH: Final = 1024
 
 
-class AgentSkillsIndexEntry(BaseModel):
+class AgentSkillsIndexEntry(LiteLLMBaseModel):
     name: str
     type: Literal["archive"]
     description: str
@@ -20,6 +22,6 @@ class AgentSkillsIndexEntry(BaseModel):
     digest: str
 
 
-class AgentSkillsIndex(BaseModel):
+class AgentSkillsIndex(LiteLLMBaseModel):
     discovery_schema: str = Field(default=AGENT_SKILLS_DISCOVERY_SCHEMA_URL, alias="$schema")
     skills: tuple[AgentSkillsIndexEntry, ...]
diff --git a/litellm/types/proxy/discovery_endpoints/ui_discovery_endpoints.py b/litellm/types/proxy/discovery_endpoints/ui_discovery_endpoints.py
index 4e0faabc132..c30fb7caf2e 100644
--- a/litellm/types/proxy/discovery_endpoints/ui_discovery_endpoints.py
+++ b/litellm/types/proxy/discovery_endpoints/ui_discovery_endpoints.py
@@ -1,9 +1,10 @@
-from pydantic import BaseModel
+from pydantic import Field
 
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.control_plane_endpoints import WorkerRegistryEntry
 
 
-class UiDiscoveryEndpoints(BaseModel):
+class UiDiscoveryEndpoints(LiteLLMBaseModel):
     server_root_path: str
     proxy_base_url: str | None
     auto_redirect_to_sso: bool
@@ -12,4 +13,4 @@ class UiDiscoveryEndpoints(BaseModel):
     hide_default_credentials_hint: bool = False
     is_control_plane: bool = False
     mcp_stdio_enabled: bool = False
-    workers: list[WorkerRegistryEntry] = []
+    workers: list[WorkerRegistryEntry] = Field(default=[])
diff --git a/litellm/types/proxy/gateway_requests.py b/litellm/types/proxy/gateway_requests.py
index f0abeb3c950..77607899dde 100644
--- a/litellm/types/proxy/gateway_requests.py
+++ b/litellm/types/proxy/gateway_requests.py
@@ -4,7 +4,7 @@ from collections.abc import Mapping
 from dataclasses import dataclass
 from typing import TypeAlias
 
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
 @dataclass(frozen=True, slots=True)
@@ -29,20 +29,20 @@ class GatewayRequestCounts:
 GatewayRequestSnapshot: TypeAlias = Mapping[GatewayRequestKey, GatewayRequestCounts]
 
 
-class GatewayRequestBreakdownEntry(BaseModel):
+class GatewayRequestBreakdownEntry(LiteLLMBaseModel):
     category: str
     route: str
     successful_requests: int = 0
     failed_requests: int = 0
 
 
-class GatewayRequestDailyEntry(BaseModel):
+class GatewayRequestDailyEntry(LiteLLMBaseModel):
     date: str
     successful_requests: int = 0
     failed_requests: int = 0
 
 
-class GatewayRequestActivityResponse(BaseModel):
+class GatewayRequestActivityResponse(LiteLLMBaseModel):
     """Response for GET /gateway/daily/activity."""
 
     total_successful_requests: int = 0
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/azure/azure_text_moderation.py b/litellm/types/proxy/guardrails/guardrail_hooks/azure/azure_text_moderation.py
index 83b8d281100..90662e84ed5 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/azure/azure_text_moderation.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/azure/azure_text_moderation.py
@@ -1,8 +1,10 @@
 from typing import Any, Final, Literal
 
-from pydantic import BaseModel, Field
+from pydantic import Field
 from typing_extensions import Required, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 from ..base import GuardrailConfigModel
 from .base import AzureContentSafetyConfigModel
 
@@ -42,7 +44,7 @@ class AzureTextModerationGuardrailResponse(TypedDict):
 AzureHarmCategories = Literal["Hate", "SelfHarm", "Sexual", "Violence"]
 
 
-class AzureTextModerationOptionalParams(BaseModel):
+class AzureTextModerationOptionalParams(LiteLLMBaseModel):
     severity_threshold: int | None = Field(
         default=None,
         description="Severity threshold for the Azure Content Safety Text Moderation guardrail across all categories",
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/azure/base.py b/litellm/types/proxy/guardrails/guardrail_hooks/azure/base.py
index c72888f33e1..cc09e2b679d 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/azure/base.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/azure/base.py
@@ -1,7 +1,9 @@
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class AzureContentSafetyConfigModel(BaseModel):
+class AzureContentSafetyConfigModel(LiteLLMBaseModel):
     """Configuration parameters for the Azure Content Safety Prompt Shield guardrail"""
 
     api_key: str | None = Field(
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/base.py b/litellm/types/proxy/guardrails/guardrail_hooks/base.py
index d965ad361bc..a3c5e74d701 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/base.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/base.py
@@ -3,10 +3,12 @@ from typing import Generic, TypeVar
 
 from pydantic import BaseModel, Field
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 T = TypeVar("T", bound=BaseModel)
 
 
-class GuardrailConfigModel(BaseModel, Generic[T], ABC):
+class GuardrailConfigModel(LiteLLMBaseModel, Generic[T], ABC):
     """Base model for guardrail configuration"""
 
     optional_params: T | None = Field(
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/cisco_ai_defense.py b/litellm/types/proxy/guardrails/guardrail_hooks/cisco_ai_defense.py
index 4d6d28dd07b..46f1bd33e47 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/cisco_ai_defense.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/cisco_ai_defense.py
@@ -4,7 +4,9 @@ Cisco AI Defense Guardrail Config Model
 
 from typing import Literal
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .base import GuardrailConfigModel
 
@@ -30,7 +32,7 @@ CISCO_AI_DEFENSE_RULE_NAMES = Literal[
 CISCO_AI_DEFENSE_INSPECTION_TYPE = Literal["chat", "mcp"]
 
 
-class CiscoAIDefenseRule(BaseModel):
+class CiscoAIDefenseRule(LiteLLMBaseModel):
     """A single rule to enable for Cisco AI Defense inspection."""
 
     rule_name: CISCO_AI_DEFENSE_RULE_NAMES = Field(
@@ -45,7 +47,7 @@ class CiscoAIDefenseRule(BaseModel):
     )
 
 
-class CiscoAIDefenseGuardrailConfigModelOptionalParams(BaseModel):
+class CiscoAIDefenseGuardrailConfigModelOptionalParams(LiteLLMBaseModel):
     """Optional parameters for the Cisco AI Defense guardrail."""
 
     model_config = ConfigDict(extra="allow")
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/compresr.py b/litellm/types/proxy/guardrails/guardrail_hooks/compresr.py
index 9efc2fbff55..6598f1de62b 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/compresr.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/compresr.py
@@ -1,11 +1,13 @@
 from typing import Any, Literal
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .base import GuardrailConfigModel
 
 
-class CompresrGuardrailOptionalParams(BaseModel):
+class CompresrGuardrailOptionalParams(LiteLLMBaseModel):
     """Optional tuning knobs for the Compresr guardrail."""
 
     target_compression_ratio: float | None = Field(
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/conduct.py b/litellm/types/proxy/guardrails/guardrail_hooks/conduct.py
index fbff4363351..3c541e289e1 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/conduct.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/conduct.py
@@ -2,12 +2,14 @@ from __future__ import annotations
 
 from typing import Literal
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .base import GuardrailConfigModel
 
 
-class ConductGuardrailConfigModelOptionalParams(BaseModel):
+class ConductGuardrailConfigModelOptionalParams(LiteLLMBaseModel):
     workspace_id: str | None = Field(
         default=None,
         description="Conduct workspace id, sent as the X-Workspace-Id header. Env: CONDUCT_WORKSPACE_ID.",
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/crowdstrike_aidr.py b/litellm/types/proxy/guardrails/guardrail_hooks/crowdstrike_aidr.py
index df1caab6af6..66e1e12e99c 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/crowdstrike_aidr.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/crowdstrike_aidr.py
@@ -1,9 +1,11 @@
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .base import GuardrailConfigModel
 
 
-class CrowdStrikeAIDRGuardrailConfigModelOptionalParams(BaseModel):
+class CrowdStrikeAIDRGuardrailConfigModelOptionalParams(LiteLLMBaseModel):
     streaming_buffer_until_moderated: bool | None = Field(
         default=None,
         description="When True, withhold streamed chunks until moderation passes. Defaults to False when unset.",
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/deepkeep.py b/litellm/types/proxy/guardrails/guardrail_hooks/deepkeep.py
index 9830faffc23..67a3466919a 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/deepkeep.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/deepkeep.py
@@ -1,9 +1,11 @@
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .base import GuardrailConfigModel
 
 
-class DeepKeepGuardrailConfigModelOptionalParams(BaseModel):
+class DeepKeepGuardrailConfigModelOptionalParams(LiteLLMBaseModel):
     unreachable_fallback: str | None = Field(
         default="fail_closed",
         description=(
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/enkryptai.py b/litellm/types/proxy/guardrails/guardrail_hooks/enkryptai.py
index 9b0d448ded2..d1eba1dd4a9 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/enkryptai.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/enkryptai.py
@@ -1,8 +1,10 @@
 from typing import Any
 
-from pydantic import BaseModel, Field
+from pydantic import Field
 from typing_extensions import TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 from .base import GuardrailConfigModel
 
 # TypedDicts for EnkryptAI API Response Structure
@@ -110,7 +112,7 @@ class EnkryptAIProcessedResult(TypedDict):
 
 
 # Pydantic Config Model
-class EnkryptAIGuardrailConfigs(BaseModel):
+class EnkryptAIGuardrailConfigs(LiteLLMBaseModel):
     """Configuration parameters for the EnkryptAI guardrail"""
 
     api_key: str | None = Field(
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/generic_guardrail_api.py b/litellm/types/proxy/guardrails/guardrail_hooks/generic_guardrail_api.py
index 44e2cc2404f..49eaa154fec 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/generic_guardrail_api.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/generic_guardrail_api.py
@@ -1,9 +1,10 @@
 from collections.abc import Mapping, Sequence
 from typing import Any, Final, Literal, cast  # noqa: TID251  # JSON chat rows have no typed constructor across roles
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
 from typing_extensions import TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.llms.openai import (
     AllMessageValues,
     ChatCompletionToolCallChunk,
@@ -12,7 +13,7 @@ from litellm.types.proxy.guardrails.guardrail_hooks.base import GuardrailConfigM
 from litellm.types.utils import ChatCompletionMessageToolCall
 
 
-class GuardrailToolParam(BaseModel):
+class GuardrailToolParam(LiteLLMBaseModel):
     """A tool forwarded verbatim to the guardrail for inspection.
 
     Built-in tools (code_interpreter, file_search, ...) have no ``function`` block
@@ -35,7 +36,7 @@ class GenericGuardrailAPIMetadata(TypedDict, total=False):
     user_api_key_org_id: str | None
 
 
-class GenericGuardrailAPIOptionalParams(BaseModel):
+class GenericGuardrailAPIOptionalParams(LiteLLMBaseModel):
     """Optional parameters for the Generic Guardrail API"""
 
     additional_provider_specific_params: dict[str, Any] | None = Field(
@@ -119,7 +120,7 @@ class GenericGuardrailAPIConfigModel(
         return "Generic Guardrail API"
 
 
-class GenericGuardrailAPIRequest(BaseModel):
+class GenericGuardrailAPIRequest(LiteLLMBaseModel):
     """Request model for the Generic Guardrail API"""
 
     input_type: Literal["request", "response"]
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/grayswan.py b/litellm/types/proxy/guardrails/guardrail_hooks/grayswan.py
index 796ff2818a4..9c2cb1e517e 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/grayswan.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/grayswan.py
@@ -1,11 +1,13 @@
 """Gray Swan guardrail configuration models."""
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .base import GuardrailConfigModel
 
 
-class GraySwanGuardrailConfigModelOptionalParams(BaseModel):
+class GraySwanGuardrailConfigModelOptionalParams(LiteLLMBaseModel):
     """Optional parameters for the Gray Swan guardrail."""
 
     on_flagged_action: str | None = Field(
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/ibm/base.py b/litellm/types/proxy/guardrails/guardrail_hooks/ibm/base.py
index 22f6b2fd7dc..e695ab0b4b8 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/ibm/base.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/ibm/base.py
@@ -1,7 +1,9 @@
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class IBMGuardrailsBaseConfigModel(BaseModel):
+class IBMGuardrailsBaseConfigModel(LiteLLMBaseModel):
     """Base configuration parameters for IBM Guardrails"""
 
     auth_token: str | None = Field(
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/ibm/ibm_detector.py b/litellm/types/proxy/guardrails/guardrail_hooks/ibm/ibm_detector.py
index 5226d5fe6de..b58d1bff8a2 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/ibm/ibm_detector.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/ibm/ibm_detector.py
@@ -1,8 +1,10 @@
 from typing import Any
 
-from pydantic import BaseModel, Field
+from pydantic import Field
 from typing_extensions import TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 from ..base import GuardrailConfigModel
 from .base import IBMGuardrailsBaseConfigModel
 
@@ -52,7 +54,7 @@ class IBMDetectorResponseOrchestrator(TypedDict):
 # Pydantic Config Models
 
 
-class IBMDetectorOptionalParams(BaseModel):
+class IBMDetectorOptionalParams(LiteLLMBaseModel):
     """Optional parameters for IBM Detector guardrail"""
 
     detector_params: dict[str, Any] | None = Field(
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/lasso.py b/litellm/types/proxy/guardrails/guardrail_hooks/lasso.py
index e562e79dd45..77b13ad5f5d 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/lasso.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/lasso.py
@@ -1,9 +1,11 @@
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .base import GuardrailConfigModel
 
 
-class LassoGuardrailConfigModelOptionalParams(BaseModel):
+class LassoGuardrailConfigModelOptionalParams(LiteLLMBaseModel):
     user_id: str | None = Field(
         default=None,
         description="The user ID for the Lasso guardrail. If not provided, the `LASSO_USER_ID` environment variable is checked.",
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/pangea.py b/litellm/types/proxy/guardrails/guardrail_hooks/pangea.py
index e9a01810151..55c65158e30 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/pangea.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/pangea.py
@@ -1,9 +1,11 @@
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .base import GuardrailConfigModel
 
 
-class PangeaGuardrailConfigModelOptionalParams(BaseModel):
+class PangeaGuardrailConfigModelOptionalParams(LiteLLMBaseModel):
     pangea_input_recipe: str | None = Field(
         default=None,
         description="The Pangea input recipe for the Pangea guardrail. Used for pre-call hook.",
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/pillar.py b/litellm/types/proxy/guardrails/guardrail_hooks/pillar.py
index 248f3f6d2d6..e4c7ca0f589 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/pillar.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/pillar.py
@@ -2,12 +2,14 @@
 Pillar Security Guardrail Config Model
 """
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .base import GuardrailConfigModel
 
 
-class PillarGuardrailConfigModelOptionalParams(BaseModel):
+class PillarGuardrailConfigModelOptionalParams(LiteLLMBaseModel):
     """Optional parameters for the Pillar Security guardrail"""
 
     on_flagged_action: str | None = Field(
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/singulr.py b/litellm/types/proxy/guardrails/guardrail_hooks/singulr.py
index ea1e6238181..68713bf1a55 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/singulr.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/singulr.py
@@ -1,34 +1,36 @@
 from collections.abc import Mapping, Sequence
 from typing import Literal
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .base import GuardrailConfigModel
 
 
-class ContentBlock(BaseModel):
+class ContentBlock(LiteLLMBaseModel):
     type: str | None = None
     text: str | None = None
 
 
-class ToolCallFunction(BaseModel):
+class ToolCallFunction(LiteLLMBaseModel):
     name: str
     arguments: str
 
 
-class ToolCall(BaseModel):
+class ToolCall(LiteLLMBaseModel):
     id: str
     type: str = "function"
     function: ToolCallFunction
 
 
-class AssistantMessage(BaseModel):
+class AssistantMessage(LiteLLMBaseModel):
     role: Literal["assistant"] = "assistant"
     content: str | Sequence[ContentBlock] | None = None
     tool_calls: Sequence[ToolCall] | None = None
 
 
-class SingulrGuardrailPayload(BaseModel):
+class SingulrGuardrailPayload(LiteLLMBaseModel):
     correlation_id: str | None = None
     model_name: str | None = None
     model_provider_name: str | None = None
@@ -40,7 +42,7 @@ class SingulrGuardrailPayload(BaseModel):
     metadata: Mapping[str, str] | None = None
 
 
-class SingulrMcpGuardrailPayload(BaseModel):
+class SingulrMcpGuardrailPayload(LiteLLMBaseModel):
     model_name: str | None = None
     guardrail_scope: str | None = None
     tool_name: str | None = None
@@ -50,7 +52,7 @@ class SingulrMcpGuardrailPayload(BaseModel):
     metadata: Mapping[str, str] | None = None
 
 
-class SingulrGuardrailResponse(BaseModel):
+class SingulrGuardrailResponse(LiteLLMBaseModel):
     """Response returned by the Singulr guardrail API."""
 
     should_block: bool = False
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/straiker.py b/litellm/types/proxy/guardrails/guardrail_hooks/straiker.py
index 18a4608c46c..e90bd3a1796 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/straiker.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/straiker.py
@@ -2,7 +2,9 @@ from __future__ import annotations
 
 from typing import Final, Literal
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .base import GuardrailConfigModel
 
@@ -13,18 +15,18 @@ StraikerWebhookAction = Literal["NONE", "BLOCKED", "GUARDRAIL_INTERVENED"]
 STRAIKER_WEBHOOK_SCHEMA_VERSION: Final = "1"
 
 
-class StraikerWebhookStream(BaseModel):
+class StraikerWebhookStream(LiteLLMBaseModel):
     phase: StraikerWebhookStreamPhase = "none"
     index: int | None = None
 
 
-class StraikerWebhookEvent(BaseModel):
+class StraikerWebhookEvent(LiteLLMBaseModel):
     type: StraikerWebhookEventType
     id: str
     stream: StraikerWebhookStream = Field(default_factory=StraikerWebhookStream)
 
 
-class StraikerWebhookContent(BaseModel):
+class StraikerWebhookContent(LiteLLMBaseModel):
     model_config = ConfigDict(extra="ignore")
 
     texts: list[str] = Field(default_factory=list)
@@ -35,12 +37,12 @@ class StraikerWebhookContent(BaseModel):
     finish_reason: str | None = None
 
 
-class StraikerWebhookUsage(BaseModel):
+class StraikerWebhookUsage(LiteLLMBaseModel):
     input_tokens: int | None = None
     output_tokens: int | None = None
 
 
-class StraikerWebhookContext(BaseModel):
+class StraikerWebhookContext(LiteLLMBaseModel):
     call_surface: str
     mode: list[str] | None = None
     model: str | None = None
@@ -52,7 +54,7 @@ class StraikerWebhookContext(BaseModel):
     litellm_version: str | None = None
 
 
-class StraikerWebhookIdentity(BaseModel):
+class StraikerWebhookIdentity(LiteLLMBaseModel):
     litellm_key: str | None = None
     litellm_team: str | None = None
     litellm_user_id: str | None = None
@@ -61,12 +63,12 @@ class StraikerWebhookIdentity(BaseModel):
     end_user_id: str | None = None
 
 
-class StraikerWebhookApplication(BaseModel):
+class StraikerWebhookApplication(LiteLLMBaseModel):
     source: str
     name: str | None = None
 
 
-class StraikerWebhookRequest(BaseModel):
+class StraikerWebhookRequest(LiteLLMBaseModel):
     schema_version: str = STRAIKER_WEBHOOK_SCHEMA_VERSION
     event: StraikerWebhookEvent
     request: StraikerWebhookContent
@@ -78,7 +80,7 @@ class StraikerWebhookRequest(BaseModel):
     metadata: dict[str, object] | None = None
 
 
-class StraikerWebhookResponse(BaseModel):
+class StraikerWebhookResponse(LiteLLMBaseModel):
     model_config = ConfigDict(extra="allow")
 
     action: StraikerWebhookAction = "NONE"
@@ -91,7 +93,7 @@ class StraikerWebhookResponse(BaseModel):
     turn_id: str | None = Field(default=None, alias="turnId")
 
 
-class StraikerGuardrailConfigModelOptionalParams(BaseModel):
+class StraikerGuardrailConfigModelOptionalParams(LiteLLMBaseModel):
     timeout: float | None = Field(
         default=5.0,
         gt=0.0,
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/tool_permission.py b/litellm/types/proxy/guardrails/guardrail_hooks/tool_permission.py
index a31d198d605..6f521d1fd1b 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/tool_permission.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/tool_permission.py
@@ -1,12 +1,14 @@
 # Tool Permission Guardrail Type Definitions
 from typing import Final, Literal
 
-from pydantic import BaseModel, Field, field_validator, model_validator
+from pydantic import Field, field_validator, model_validator
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .base import GuardrailConfigModel
 
 
-class ToolPermissionRule(BaseModel):
+class ToolPermissionRule(LiteLLMBaseModel):
     """
     A rule defining permission for a specific tool or tool pattern
     """
@@ -53,7 +55,7 @@ class ToolPermissionRule(BaseModel):
         return self
 
 
-class ToolResult(BaseModel):
+class ToolResult(LiteLLMBaseModel):
     """
     Represents a tool_result block to be added to the response
     """
@@ -64,7 +66,7 @@ class ToolResult(BaseModel):
     is_error: bool = Field(default=True, description="Whether this is an error result")
 
 
-class PermissionError(BaseModel):
+class PermissionError(LiteLLMBaseModel):
     """
     Error information for permission denial
     """
diff --git a/litellm/types/proxy/guardrails/guardrail_hooks/typesafe.py b/litellm/types/proxy/guardrails/guardrail_hooks/typesafe.py
index 59482d2e190..d79d65760ce 100644
--- a/litellm/types/proxy/guardrails/guardrail_hooks/typesafe.py
+++ b/litellm/types/proxy/guardrails/guardrail_hooks/typesafe.py
@@ -1,11 +1,13 @@
 from typing import Literal
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from .base import GuardrailConfigModel
 
 
-class TypeSafeGuardrailOptionalParams(BaseModel):
+class TypeSafeGuardrailOptionalParams(LiteLLMBaseModel):
     """Optional tuning knobs for the TypeSafe (Jev) compaction guardrail."""
 
     relevance_threshold: float | None = Field(
diff --git a/litellm/types/proxy/management_endpoints/common_daily_activity.py b/litellm/types/proxy/management_endpoints/common_daily_activity.py
index 37804032569..9f7f666f494 100644
--- a/litellm/types/proxy/management_endpoints/common_daily_activity.py
+++ b/litellm/types/proxy/management_endpoints/common_daily_activity.py
@@ -2,9 +2,11 @@ from datetime import date
 from enum import Enum
 from typing import Any
 
-from pydantic import BaseModel, Field
+from pydantic import Field
 from typing_extensions import TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 
 class GroupByDimension(str, Enum):
     DATE = "date"
@@ -16,7 +18,7 @@ class GroupByDimension(str, Enum):
     PROVIDER = "custom_llm_provider"
 
 
-class SpendMetrics(BaseModel):
+class SpendMetrics(LiteLLMBaseModel):
     spend: float = Field(default=0.0)
     flat_cost: float = Field(default=0.0)
     prompt_tokens: int = Field(default=0)
@@ -36,11 +38,11 @@ class SpendMetrics(BaseModel):
     timed_requests: int = Field(default=0)
 
 
-class MetricBase(BaseModel):
+class MetricBase(LiteLLMBaseModel):
     metrics: SpendMetrics
 
 
-class KeyMetadata(BaseModel):
+class KeyMetadata(LiteLLMBaseModel):
     """Metadata for a key"""
 
     key_alias: str | None = None
@@ -62,7 +64,7 @@ class MetricWithMetadata(MetricBase):
     api_key_breakdown: dict[str, KeyMetricWithMetadata] = Field(default_factory=dict)  # api_key -> {metrics, metadata}
 
 
-class BreakdownMetrics(BaseModel):
+class BreakdownMetrics(LiteLLMBaseModel):
     """Breakdown of spend by different dimensions"""
 
     mcp_servers: dict[str, MetricWithMetadata] = Field(default_factory=dict)  # mcp_server -> {metrics, metadata}
@@ -74,13 +76,13 @@ class BreakdownMetrics(BaseModel):
     entities: dict[str, MetricWithMetadata] = Field(default_factory=dict)  # entity -> {metrics, metadata}
 
 
-class DailySpendData(BaseModel):
+class DailySpendData(LiteLLMBaseModel):
     date: date
     metrics: SpendMetrics
     breakdown: BreakdownMetrics = Field(default_factory=BreakdownMetrics)
 
 
-class DailySpendMetadata(BaseModel):
+class DailySpendMetadata(LiteLLMBaseModel):
     total_spend: float = Field(default=0.0)
     total_flat_cost: float = Field(default=0.0)
     total_prompt_tokens: int = Field(default=0)
@@ -119,18 +121,18 @@ class DailySpendMetadata(BaseModel):
     )
 
 
-class SpendAnalyticsPaginatedResponse(BaseModel):
+class SpendAnalyticsPaginatedResponse(LiteLLMBaseModel):
     results: list[DailySpendData]
     metadata: DailySpendMetadata = Field(default_factory=DailySpendMetadata)
 
 
-class KeyActivityRow(BaseModel):
+class KeyActivityRow(LiteLLMBaseModel):
     api_key: str
     metrics: SpendMetrics
     metadata: KeyMetadata
 
 
-class KeySpendMetrics(BaseModel):
+class KeySpendMetrics(LiteLLMBaseModel):
     spend: float = 0.0
     prompt_tokens: int = 0
     completion_tokens: int = 0
@@ -142,34 +144,34 @@ class KeySpendMetrics(BaseModel):
     cache_creation_input_tokens: int = 0
 
 
-class KeySpendActivityRow(BaseModel):
+class KeySpendActivityRow(LiteLLMBaseModel):
     api_key: str
     metrics: KeySpendMetrics
     metadata: KeyMetadata
 
 
-class DailyActivityKeySearchResponse(BaseModel):
+class DailyActivityKeySearchResponse(LiteLLMBaseModel):
     api_keys: list[KeyActivityRow]
 
 
-class DailyActivityKeyPageResponse(BaseModel):
+class DailyActivityKeyPageResponse(LiteLLMBaseModel):
     api_keys: list[KeySpendActivityRow]
     total_api_keys: int
     offset: int
     limit: int
 
 
-class ModelTopKeysResponse(BaseModel):
+class ModelTopKeysResponse(LiteLLMBaseModel):
     model: str
     by_model_group: bool
     api_keys: list[KeySpendActivityRow]
 
 
-class CacheLeakageKeysResponse(BaseModel):
+class CacheLeakageKeysResponse(LiteLLMBaseModel):
     api_keys: list[KeySpendActivityRow]
 
 
-class LiteLLM_DailyUserSpend(BaseModel):
+class LiteLLM_DailyUserSpend(LiteLLMBaseModel):
     id: str
     user_id: str
     date: str
diff --git a/litellm/types/proxy/management_endpoints/config_overrides.py b/litellm/types/proxy/management_endpoints/config_overrides.py
index 2e0fce08545..c82e616cf9a 100644
--- a/litellm/types/proxy/management_endpoints/config_overrides.py
+++ b/litellm/types/proxy/management_endpoints/config_overrides.py
@@ -1,9 +1,11 @@
 from typing import Any
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class HashicorpVaultConfig(BaseModel):
+class HashicorpVaultConfig(LiteLLMBaseModel):
     """Configuration for Hashicorp Vault secret manager integration."""
 
     vault_addr: str | None = Field(
@@ -60,7 +62,7 @@ class HashicorpVaultConfig(BaseModel):
     )
 
 
-class CyberArkConfig(BaseModel):
+class CyberArkConfig(LiteLLMBaseModel):
     """Configuration for CyberArk Conjur secret manager integration."""
 
     cyberark_api_base: str | None = Field(
@@ -97,7 +99,7 @@ class CyberArkConfig(BaseModel):
     )
 
 
-class ConfigOverrideSettingsResponse(BaseModel):
+class ConfigOverrideSettingsResponse(LiteLLMBaseModel):
     """Response model for config override settings GET endpoints."""
 
     config_type: str = Field(description="The type of config override")
diff --git a/litellm/types/proxy/management_endpoints/customer_endpoints.py b/litellm/types/proxy/management_endpoints/customer_endpoints.py
index 73fbf711556..c1c4c496ede 100644
--- a/litellm/types/proxy/management_endpoints/customer_endpoints.py
+++ b/litellm/types/proxy/management_endpoints/customer_endpoints.py
@@ -1,7 +1,8 @@
-from pydantic import BaseModel, Field
+from pydantic import Field
 
 from litellm.models.budget import LiteLLM_BudgetTableFull
 from litellm.models.end_user import LiteLLM_EndUserTable
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
 class CustomerResponse(LiteLLM_EndUserTable):
@@ -15,14 +16,14 @@ class CustomerResponse(LiteLLM_EndUserTable):
     litellm_budget_table: LiteLLM_BudgetTableFull | None = None  # pyright: ignore
 
 
-class BlockUsersResponse(BaseModel):
+class BlockUsersResponse(LiteLLMBaseModel):
     blocked_users: list[LiteLLM_EndUserTable]
 
 
-class UnblockUsersResponse(BaseModel):
+class UnblockUsersResponse(LiteLLMBaseModel):
     blocked_users: list[str] = Field(description="User IDs that remain blocked after this unblock call")
 
 
-class DeleteCustomersResponse(BaseModel):
+class DeleteCustomersResponse(LiteLLMBaseModel):
     deleted_customers: int
     message: str
diff --git a/litellm/types/proxy/management_endpoints/internal_user_endpoints.py b/litellm/types/proxy/management_endpoints/internal_user_endpoints.py
index 43e3899d523..1fd3e7dca53 100644
--- a/litellm/types/proxy/management_endpoints/internal_user_endpoints.py
+++ b/litellm/types/proxy/management_endpoints/internal_user_endpoints.py
@@ -1,7 +1,7 @@
 from collections.abc import Mapping, Sequence
 from typing import Any, Final, Literal
 
-from pydantic import BaseModel, ConfigDict, Field, field_validator
+from pydantic import ConfigDict, Field, field_validator
 from typing_extensions import ReadOnly, TypedDict
 
 from litellm.proxy._types import (
@@ -10,6 +10,7 @@ from litellm.proxy._types import (
     UpdateUserRequest,
     UpdateUserRequestNoUserIDorEmail,
 )
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.management_endpoints.management_v1 import ResourceResponse
 
 MAX_BULK_DELETE_USERS: Final = 500
@@ -28,7 +29,7 @@ class UserSearchWhere(TypedDict):
     OR: ReadOnly[tuple[Mapping[Literal["user_id", "user_email"], InsensitiveContains], ...]]
 
 
-class UserListResponse(BaseModel):
+class UserListResponse(LiteLLMBaseModel):
     """
     Response model for the user list endpoint
     """
@@ -40,7 +41,7 @@ class UserListResponse(BaseModel):
     total_pages: int
 
 
-class BulkUpdateUserRequest(BaseModel):
+class BulkUpdateUserRequest(LiteLLMBaseModel):
     """Request for bulk user updates"""
 
     users: list[UpdateUserRequest] | None = None  # List of specific user update requests
@@ -72,7 +73,7 @@ class BulkUpdateUserRequest(BaseModel):
         return v
 
 
-class UserUpdateResult(BaseModel):
+class UserUpdateResult(LiteLLMBaseModel):
     """Result of a single user update operation"""
 
     user_id: str | None = None
@@ -82,7 +83,7 @@ class UserUpdateResult(BaseModel):
     updated_user: dict[str, Any] | None = None
 
 
-class BulkUpdateUserResponse(BaseModel):
+class BulkUpdateUserResponse(LiteLLMBaseModel):
     """Response for bulk user update operations"""
 
     results: list[UserUpdateResult]
@@ -91,7 +92,7 @@ class BulkUpdateUserResponse(BaseModel):
     failed_updates: int
 
 
-class BulkDeleteUserRequest(BaseModel):
+class BulkDeleteUserRequest(LiteLLMBaseModel):
     """Body of `POST /management/v1/users/bulk_delete`."""
 
     model_config = ConfigDict(extra="forbid")
@@ -99,7 +100,7 @@ class BulkDeleteUserRequest(BaseModel):
     user_ids: tuple[str, ...] = Field(min_length=1, max_length=MAX_BULK_DELETE_USERS)
 
 
-class UserDeleteResult(BaseModel):
+class UserDeleteResult(LiteLLMBaseModel):
     """Outcome for one requested user, in request order. `teams_removed` lists the teams the user left."""
 
     user_id: str
@@ -129,13 +130,13 @@ class BulkNewUserItem(NewUserRequest):
         return value
 
 
-class BulkNewUserRequest(BaseModel):
+class BulkNewUserRequest(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid")
 
     users: Sequence[BulkNewUserItem] = Field(min_length=1, max_length=MAX_BULK_NEW_USERS)
 
 
-class UserCreateResult(BaseModel):
+class UserCreateResult(LiteLLMBaseModel):
     """Outcome for one row of `POST /management/v1/users/bulk`. `teams` lists the teams the user was actually
     added to."""
 
@@ -147,13 +148,13 @@ class UserCreateResult(BaseModel):
     error: str | None = None
 
 
-class BulkNewUserMeta(BaseModel):
+class BulkNewUserMeta(LiteLLMBaseModel):
     total_requested: int
     created: int
     failed: int
 
 
-class BulkNewUserResponse(BaseModel):
+class BulkNewUserResponse(LiteLLMBaseModel):
     """`data` holds one result per input row, in input order."""
 
     data: tuple[UserCreateResult, ...]
diff --git a/litellm/types/proxy/management_endpoints/key_management_endpoints.py b/litellm/types/proxy/management_endpoints/key_management_endpoints.py
index 001bc3c0d51..05b63100ea4 100644
--- a/litellm/types/proxy/management_endpoints/key_management_endpoints.py
+++ b/litellm/types/proxy/management_endpoints/key_management_endpoints.py
@@ -1,7 +1,7 @@
 from datetime import datetime
 from typing import Any, Final, Literal, TypeAlias
 
-from pydantic import BaseModel, ConfigDict, model_validator
+from pydantic import ConfigDict, model_validator
 from typing_extensions import ReadOnly, TypedDict
 
 from litellm.models.verification_token import LiteLLM_VerificationToken
@@ -11,7 +11,7 @@ from litellm.proxy._types import (
     RegenerateKeyRequest,
     UpdateKeyRequest,
 )
-from litellm.types.llms.base import LiteLLMPydanticObjectBase
+from litellm.types.llms.base import LiteLLMBaseModel, LiteLLMPydanticObjectBase
 from litellm.types.proxy.management_endpoints.internal_user_endpoints import InsensitiveContains
 
 
@@ -29,7 +29,7 @@ class KeySearchWhere(TypedDict):
     OR: ReadOnly[tuple[KeyTokenWhere, KeyAliasContainsWhere]]
 
 
-class BulkUpdateKeyRequestItem(BaseModel):
+class BulkUpdateKeyRequestItem(LiteLLMBaseModel):
     """One /key/bulk_update item; only the fields it carries are written."""
 
     key: str  # Key identifier (token)
@@ -40,20 +40,20 @@ class BulkUpdateKeyRequestItem(BaseModel):
     object_permission: LiteLLM_ObjectPermissionBase | None = None
 
 
-class BulkUpdateKeyRequest(BaseModel):
+class BulkUpdateKeyRequest(LiteLLMBaseModel):
     """Request for bulk key updates"""
 
     keys: list[BulkUpdateKeyRequestItem]
 
 
-class SuccessfulKeyUpdate(BaseModel):
+class SuccessfulKeyUpdate(LiteLLMBaseModel):
     """Successfully updated key with its updated information"""
 
     key: str
     key_info: dict[str, Any]
 
 
-class FailedKeyUpdate(BaseModel):
+class FailedKeyUpdate(LiteLLMBaseModel):
     """Failed key update with reason"""
 
     key: str
@@ -61,7 +61,7 @@ class FailedKeyUpdate(BaseModel):
     failed_reason: str
 
 
-class BulkUpdateKeyResponse(BaseModel):
+class BulkUpdateKeyResponse(LiteLLMBaseModel):
     """Response for bulk key update operations"""
 
     total_requested: int
@@ -69,7 +69,7 @@ class BulkUpdateKeyResponse(BaseModel):
     failed_updates: list[FailedKeyUpdate]
 
 
-class KeyUpdateFields(BaseModel):
+class KeyUpdateFields(LiteLLMBaseModel):
     """Allowlist of bulk-broadcastable fields for /team/key/bulk_update; `extra="forbid"` blocks RBAC/ownership/scope mutations even by team admins."""
 
     model_config = ConfigDict(extra="forbid", protected_namespaces=())
@@ -116,7 +116,7 @@ class KeyUpdateFields(BaseModel):
         return self
 
 
-class BulkUpdateTeamKeysRequest(BaseModel):
+class BulkUpdateTeamKeysRequest(LiteLLMBaseModel):
     """Apply one update payload to many keys inside a team; provide either `key_ids` or `all_keys_in_team=True`."""
 
     team_id: str
diff --git a/litellm/types/proxy/management_endpoints/management_v1.py b/litellm/types/proxy/management_endpoints/management_v1.py
index c23d0ecfb54..6fd12e00381 100644
--- a/litellm/types/proxy/management_endpoints/management_v1.py
+++ b/litellm/types/proxy/management_endpoints/management_v1.py
@@ -3,12 +3,14 @@
 from collections.abc import Sequence
 from typing import Generic, TypeVar
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 TOut = TypeVar("TOut")
 
 
-class ProblemDetail(BaseModel):
+class ProblemDetail(LiteLLMBaseModel):
     """RFC 9457 problem details, served as `application/problem+json`."""
 
     type: str
@@ -18,7 +20,7 @@ class ProblemDetail(BaseModel):
     allowed: list[str] | None = None
 
 
-class PageLinks(BaseModel):
+class PageLinks(LiteLLMBaseModel):
     """Hypermedia for a paginated list. No `first`/`last`: without a total count the last page is unknown."""
 
     model_config = ConfigDict(populate_by_name=True)
@@ -28,7 +30,7 @@ class PageLinks(BaseModel):
     next: str | None = None
 
 
-class PageMeta(BaseModel):
+class PageMeta(LiteLLMBaseModel):
     """`has_more` rather than `total_count`, which would need a COUNT(*) over the whole match set per keystroke."""
 
     page: int
@@ -36,7 +38,7 @@ class PageMeta(BaseModel):
     has_more: bool
 
 
-class FacetListResponse(BaseModel):
+class FacetListResponse(LiteLLMBaseModel):
     """The distinct values one column takes over a filtered query. `data` holds bare values, not entity rows."""
 
     data: Sequence[str]
@@ -44,7 +46,7 @@ class FacetListResponse(BaseModel):
     links: PageLinks
 
 
-class ListMeta(BaseModel):
+class ListMeta(LiteLLMBaseModel):
     """Page-mode counterpart to `PageMeta`: an entity list pays for the COUNT(*) so the table can show a page count."""
 
     total_count: int
@@ -53,7 +55,7 @@ class ListMeta(BaseModel):
     total_pages: int
 
 
-class ListLinks(BaseModel):
+class ListLinks(LiteLLMBaseModel):
     """Page-mode counterpart to `PageLinks`. `first`/`last` are knowable here because the total count is."""
 
     model_config = ConfigDict(populate_by_name=True)
@@ -65,13 +67,13 @@ class ListLinks(BaseModel):
     last: str
 
 
-class ResourceResponse(BaseModel, Generic[TOut]):
+class ResourceResponse(LiteLLMBaseModel, Generic[TOut]):
     """Envelope for a single resource or an action's result: `{data: ...}`, no `meta` or `links`."""
 
     data: TOut
 
 
-class ListResponse(BaseModel, Generic[TOut]):
+class ListResponse(LiteLLMBaseModel, Generic[TOut]):
     """Rows stay flat: JSON:API's `{type, id, attributes}` wrapper is a deliberate deviation, so every
     dashboard column accessor would otherwise have to go through `.attributes`."""
 
diff --git a/litellm/types/proxy/management_endpoints/model_management_endpoints.py b/litellm/types/proxy/management_endpoints/model_management_endpoints.py
index 8315ac0d4d2..3cc85a9427e 100644
--- a/litellm/types/proxy/management_endpoints/model_management_endpoints.py
+++ b/litellm/types/proxy/management_endpoints/model_management_endpoints.py
@@ -1,7 +1,9 @@
 from datetime import datetime
 from typing import Any
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 from ...router import ModelGroupInfo
 
@@ -13,14 +15,14 @@ class ModelGroupInfoProxy(ModelGroupInfo):
     health_checked_at: str | None = Field(default=None)
 
 
-class UpdateUsefulLinksRequest(BaseModel):
+class UpdateUsefulLinksRequest(LiteLLMBaseModel):
     # Supports both old format (Dict[str, str]) and new format (Dict[str, Dict[str, Any]])
     # New format: { "displayName": { "url": "...", "index": 0 } }
     # Old format: { "displayName": "url" } (for backward compatibility)
     useful_links: dict[str, str | dict[str, Any]]
 
 
-class AutoRouterClassifierDefaultPromptResponse(BaseModel):
+class AutoRouterClassifierDefaultPromptResponse(LiteLLMBaseModel):
     """The built-in system prompt an auto-router's LLM classifier uses when none is configured.
 
     Served so the dashboard's prompt editor prefills the rubric the proxy actually sends, rather than
@@ -30,31 +32,31 @@ class AutoRouterClassifierDefaultPromptResponse(BaseModel):
     system_prompt: str
 
 
-class NewModelGroupRequest(BaseModel):
+class NewModelGroupRequest(LiteLLMBaseModel):
     access_group: str  # The access group name (e.g., "production-models")
     model_names: list[str] | None = None  # Existing model groups to include - tags ALL deployments for each name
     model_ids: list[str] | None = None  # Specific deployment IDs to tag (more precise than model_names)
 
 
-class NewModelGroupResponse(BaseModel):
+class NewModelGroupResponse(LiteLLMBaseModel):
     access_group: str
     model_names: list[str] | None = None
     model_ids: list[str] | None = None
     models_updated: int  # Number of models updated
 
 
-class UpdateModelGroupRequest(BaseModel):
+class UpdateModelGroupRequest(LiteLLMBaseModel):
     model_names: list[str] | None = None  # Updated list of model groups to include - tags ALL deployments for each name
     model_ids: list[str] | None = None  # Specific deployment IDs to tag (more precise than model_names)
 
 
-class DeleteModelGroupResponse(BaseModel):
+class DeleteModelGroupResponse(LiteLLMBaseModel):
     access_group: str
     models_updated: int  # Number of deployments where the access group was removed
     message: str
 
 
-class AccessGroupBudget(BaseModel):
+class AccessGroupBudget(LiteLLMBaseModel):
     budget_id: str
     max_budget: float | None = None
     soft_budget: float | None = None
@@ -62,7 +64,7 @@ class AccessGroupBudget(BaseModel):
     budget_reset_at: datetime | None = None
 
 
-class AccessGroupBudgetRequest(BaseModel):
+class AccessGroupBudgetRequest(LiteLLMBaseModel):
     budget_id: str | None = None  # Link an existing budget instead of creating one
     max_budget: float | None = Field(default=None, ge=0)
     soft_budget: float | None = Field(default=None, ge=0)
@@ -72,19 +74,19 @@ class AccessGroupBudgetRequest(BaseModel):
     model_config = ConfigDict(extra="forbid")
 
 
-class AccessGroupBudgetResponse(BaseModel):
+class AccessGroupBudgetResponse(LiteLLMBaseModel):
     access_group: str
     spend: float  # Shared spend accrued by every key that can reach this access group
     budget: AccessGroupBudget | None = None
 
 
-class DeleteAccessGroupBudgetResponse(BaseModel):
+class DeleteAccessGroupBudgetResponse(LiteLLMBaseModel):
     access_group: str
     budget_deleted: bool  # False when the access group had no budget to begin with
     message: str
 
 
-class AccessGroupInfo(BaseModel):
+class AccessGroupInfo(LiteLLMBaseModel):
     access_group: str
     model_names: list[str]  # List of model names in this access group
     deployment_count: int  # Total number of deployments with this access group
@@ -92,5 +94,5 @@ class AccessGroupInfo(BaseModel):
     budget: AccessGroupBudget | None = None
 
 
-class ListAccessGroupsResponse(BaseModel):
+class ListAccessGroupsResponse(LiteLLMBaseModel):
     access_groups: list[AccessGroupInfo]
diff --git a/litellm/types/proxy/management_endpoints/scim_v2.py b/litellm/types/proxy/management_endpoints/scim_v2.py
index 6f2c48ab283..9b519f9af4f 100644
--- a/litellm/types/proxy/management_endpoints/scim_v2.py
+++ b/litellm/types/proxy/management_endpoints/scim_v2.py
@@ -2,7 +2,6 @@ from typing import Any, Final, Literal, Optional, Union
 
 from fastapi import HTTPException
 from pydantic import (
-    BaseModel,
     ConfigDict,
     EmailStr,
     Field,
@@ -13,6 +12,8 @@ from pydantic import (
 )
 from pydantic_core.core_schema import SerializerFunctionWrapHandler
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 SCIM_ENTERPRISE_USER_SCHEMA: Final = "urn:ietf:params:scim:schemas:extension:enterprise:2.0:User"
 SCIM_ENTERPRISE_METADATA_KEY: Final = "scim_enterprise"
 SCIM_ENTITLEMENTS_METADATA_KEY: Final = "scim_entitlements"
@@ -22,7 +23,7 @@ SCIM_MANAGED_TEAM_METADATA_KEY: Final = "scim_managed"
 SCIM_TEAM_DATA_METADATA_KEY: Final = "scim_data"
 
 
-class LiteLLM_UserScimMetadata(BaseModel):
+class LiteLLM_UserScimMetadata(LiteLLMBaseModel):
     """
     Scim metadata stored in LiteLLM_UserTable.metadata
     """
@@ -32,14 +33,14 @@ class LiteLLM_UserScimMetadata(BaseModel):
 
 
 # SCIM Resource Models
-class SCIMResource(BaseModel):
+class SCIMResource(LiteLLMBaseModel):
     schemas: list[str]
     id: str | None = None
     externalId: str | None = None
     meta: dict[str, object] | None = None
 
 
-class SCIMUserName(BaseModel):
+class SCIMUserName(LiteLLMBaseModel):
     familyName: str | None = None
     givenName: str | None = None
     formatted: str | None = None
@@ -48,19 +49,19 @@ class SCIMUserName(BaseModel):
     honorificSuffix: str | None = None
 
 
-class SCIMUserEmail(BaseModel):
+class SCIMUserEmail(LiteLLMBaseModel):
     value: EmailStr
     type: str | None = None
     primary: bool | None = None
 
 
-class SCIMUserGroup(BaseModel):
+class SCIMUserGroup(LiteLLMBaseModel):
     value: str  # Group ID
     display: str | None = None  # Group display name
     type: str | None = "direct"  # direct or indirect
 
 
-class SCIMMultiValuedAttribute(BaseModel):
+class SCIMMultiValuedAttribute(LiteLLMBaseModel):
     model_config = ConfigDict(extra="allow")
 
     value: str | None = None
@@ -84,7 +85,7 @@ SCIM_MULTI_VALUED_ATTRIBUTE_METADATA_KEYS: Final = {
 }
 
 
-class SCIMUserManager(BaseModel):
+class SCIMUserManager(LiteLLMBaseModel):
     model_config = ConfigDict(populate_by_name=True)
 
     value: str | None = None
@@ -92,7 +93,7 @@ class SCIMUserManager(BaseModel):
     ref: str | None = Field(default=None, alias="$ref")
 
 
-class SCIMEnterpriseUser(BaseModel):
+class SCIMEnterpriseUser(LiteLLMBaseModel):
     model_config = ConfigDict(populate_by_name=True)
 
     employeeNumber: str | None = None
@@ -133,7 +134,7 @@ class SCIMUser(SCIMResource):
         return dumped
 
 
-class SCIMMember(BaseModel):
+class SCIMMember(LiteLLMBaseModel):
     value: str  # User ID
     display: str | None = None  # Username or email
     type: str | None = None
@@ -152,15 +153,15 @@ class SCIMGroup(SCIMResource):
     members: list[SCIMMember] | None = None
 
 
-class SCIMPlaceholderMergeResult(BaseModel):
+class SCIMPlaceholderMergeResult(LiteLLMBaseModel):
     placeholder_user_id: str
     merged_into_user_id: str
     team_ids: tuple[str, ...]
 
 
 # SCIM List Response Models
-class SCIMListResponse(BaseModel):
-    schemas: list[str] = ["urn:ietf:params:scim:api:messages:2.0:ListResponse"]
+class SCIMListResponse(LiteLLMBaseModel):
+    schemas: list[str] = Field(default=["urn:ietf:params:scim:api:messages:2.0:ListResponse"])
     totalResults: int
     startIndex: int | None = 1
     itemsPerPage: int | None = 10
@@ -168,7 +169,7 @@ class SCIMListResponse(BaseModel):
 
 
 # SCIM PATCH Operation Models
-class SCIMPatchOperation(BaseModel):
+class SCIMPatchOperation(LiteLLMBaseModel):
     op: str
     path: str | None = None
     value: object | None = None
@@ -184,21 +185,21 @@ class SCIMPatchOperation(BaseModel):
         return v
 
 
-class SCIMPatchOp(BaseModel):
-    schemas: list[str] = ["urn:ietf:params:scim:api:messages:2.0:PatchOp"]
+class SCIMPatchOp(LiteLLMBaseModel):
+    schemas: list[str] = Field(default=["urn:ietf:params:scim:api:messages:2.0:PatchOp"])
     Operations: list[SCIMPatchOperation]
 
 
 # SCIM Service Provider Configuration Models
-class SCIMFeature(BaseModel):
+class SCIMFeature(LiteLLMBaseModel):
     supported: bool
     maxOperations: int | None = None
     maxPayloadSize: int | None = None
     maxResults: int | None = None
 
 
-class SCIMServiceProviderConfig(BaseModel):
-    schemas: list[str] = ["urn:ietf:params:scim:schemas:core:2.0:ServiceProviderConfig"]
+class SCIMServiceProviderConfig(LiteLLMBaseModel):
+    schemas: list[str] = Field(default=["urn:ietf:params:scim:schemas:core:2.0:ServiceProviderConfig"])
     patch: SCIMFeature = SCIMFeature(supported=True)
     bulk: SCIMFeature = SCIMFeature(supported=False)
     filter: SCIMFeature = SCIMFeature(supported=False)
@@ -210,7 +211,7 @@ class SCIMServiceProviderConfig(BaseModel):
 
 
 # SCIM ResourceType Models (RFC 7643 Section 6)
-class SCIMSchemaExtension(BaseModel):
+class SCIMSchemaExtension(LiteLLMBaseModel):
     model_config = ConfigDict(populate_by_name=True)
 
     schema_: str  # aliased to "schema" in serialization
@@ -222,10 +223,10 @@ class SCIMSchemaExtension(BaseModel):
         return d
 
 
-class SCIMResourceType(BaseModel):
+class SCIMResourceType(LiteLLMBaseModel):
     model_config = ConfigDict(populate_by_name=True)
 
-    schemas: list[str] = ["urn:ietf:params:scim:schemas:core:2.0:ResourceType"]
+    schemas: list[str] = Field(default=["urn:ietf:params:scim:schemas:core:2.0:ResourceType"])
     id: str
     name: str
     description: str | None = None
@@ -244,7 +245,7 @@ class SCIMResourceType(BaseModel):
 
 
 # SCIM Schema Models (RFC 7643 Section 7)
-class SCIMSchemaAttribute(BaseModel):
+class SCIMSchemaAttribute(LiteLLMBaseModel):
     name: str
     type: str
     multiValued: bool = False
@@ -262,10 +263,10 @@ class SCIMSchemaAttribute(BaseModel):
         return d
 
 
-class SCIMSchema(BaseModel):
-    schemas: list[str] = ["urn:ietf:params:scim:schemas:core:2.0:Schema"]
+class SCIMSchema(LiteLLMBaseModel):
+    schemas: list[str] = Field(default=["urn:ietf:params:scim:schemas:core:2.0:Schema"])
     id: str
     name: str
     description: str | None = None
-    attributes: list[SCIMSchemaAttribute] = []
+    attributes: list[SCIMSchemaAttribute] = Field(default=[])
     meta: dict[str, object] | None = None
diff --git a/litellm/types/proxy/management_endpoints/team_endpoints.py b/litellm/types/proxy/management_endpoints/team_endpoints.py
index 4524c47ec38..6731b826ebd 100644
--- a/litellm/types/proxy/management_endpoints/team_endpoints.py
+++ b/litellm/types/proxy/management_endpoints/team_endpoints.py
@@ -1,6 +1,6 @@
 from typing import Any, Final, Literal
 
-from pydantic import BaseModel, ConfigDict, Field, field_validator, model_validator
+from pydantic import ConfigDict, Field, field_validator, model_validator
 
 from litellm.proxy._types import (
     KeyManagementRoutes,
@@ -11,6 +11,7 @@ from litellm.proxy._types import (
     MemberDeleteRequest,
 )
 from litellm.proxy.common_utils.timezone_utils import budget_duration_error
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.management_endpoints.management_v1 import ResourceResponse
 
 TeamIdSearchMatch = Literal["exact", "prefix"]
@@ -20,13 +21,13 @@ MAX_BULK_TEAM_MEMBER_DELETES: Final = 500
 MAX_BULK_TEAM_MEMBER_BUDGET_UPDATES: Final = 500
 
 
-class GetTeamMemberPermissionsRequest(BaseModel):
+class GetTeamMemberPermissionsRequest(LiteLLMBaseModel):
     """Request to get the team member permissions for a team"""
 
     team_id: str
 
 
-class GetTeamMemberPermissionsResponse(BaseModel):
+class GetTeamMemberPermissionsResponse(LiteLLMBaseModel):
     """Response to get the team member permissions for a team"""
 
     team_id: str
@@ -34,7 +35,7 @@ class GetTeamMemberPermissionsResponse(BaseModel):
     The team id that the permissions are for
     """
 
-    team_member_permissions: list[str] | None = []
+    team_member_permissions: list[str] | None = Field(default=[])
     """
     The team member permissions currently set for the team
     """
@@ -45,14 +46,14 @@ class GetTeamMemberPermissionsResponse(BaseModel):
     """
 
 
-class UpdateTeamMemberPermissionsRequest(BaseModel):
+class UpdateTeamMemberPermissionsRequest(LiteLLMBaseModel):
     """Request to update the team member permissions for a team"""
 
     team_id: str
     team_member_permissions: list[str]
 
 
-class BulkUpdateTeamMemberPermissionsRequest(BaseModel):
+class BulkUpdateTeamMemberPermissionsRequest(LiteLLMBaseModel):
     """Request to bulk-update team member permissions across teams."""
 
     permissions: list[KeyManagementRoutes]
@@ -65,7 +66,7 @@ class BulkUpdateTeamMemberPermissionsRequest(BaseModel):
     """When True, update all teams. Mutually exclusive with team_ids."""
 
 
-class BulkUpdateTeamMemberPermissionsResponse(BaseModel):
+class BulkUpdateTeamMemberPermissionsResponse(LiteLLMBaseModel):
     """Response for bulk team member permissions update."""
 
     message: str
@@ -84,7 +85,7 @@ class TeamListItem(LiteLLM_TeamTable):
     access_group_agent_ids: list[str] | None = None
 
 
-class TeamListResponse(BaseModel):
+class TeamListResponse(LiteLLMBaseModel):
     """Response to get the list of teams"""
 
     teams: list[TeamListItem | LiteLLM_TeamTable | LiteLLM_DeletedTeamTable]
@@ -94,7 +95,7 @@ class TeamListResponse(BaseModel):
     total_pages: int
 
 
-class BulkTeamMemberAddRequest(BaseModel):
+class BulkTeamMemberAddRequest(LiteLLMBaseModel):
     """Request for bulk team member addition"""
 
     team_id: str
@@ -103,7 +104,7 @@ class BulkTeamMemberAddRequest(BaseModel):
     max_budget_in_team: float | None = None
 
 
-class TeamMemberAddResult(BaseModel):
+class TeamMemberAddResult(LiteLLMBaseModel):
     """Result of a single team member add operation"""
 
     user_id: str | None = None
@@ -114,7 +115,7 @@ class TeamMemberAddResult(BaseModel):
     updated_team_membership: dict[str, Any] | None = None
 
 
-class BulkTeamMemberAddResponse(BaseModel):
+class BulkTeamMemberAddResponse(LiteLLMBaseModel):
     """Response for bulk team member add operations"""
 
     team_id: str
@@ -137,7 +138,7 @@ class TeamMemberRef(MemberDeleteRequest):
         return self
 
 
-class BulkTeamMemberDeleteRequest(BaseModel):
+class BulkTeamMemberDeleteRequest(LiteLLMBaseModel):
     """Body of `POST /management/v1/teams/{team_id}/members/bulk_delete`."""
 
     model_config = ConfigDict(extra="forbid")
@@ -145,7 +146,7 @@ class BulkTeamMemberDeleteRequest(BaseModel):
     members: tuple[TeamMemberRef, ...] = Field(min_length=1, max_length=MAX_BULK_TEAM_MEMBER_DELETES)
 
 
-class TeamMemberDeleteResult(BaseModel):
+class TeamMemberDeleteResult(LiteLLMBaseModel):
     """Outcome for one requested member, in request order."""
 
     user_id: str | None = None
@@ -178,7 +179,7 @@ class TeamMemberBudgetPatch(TeamMemberRef):
         return value
 
 
-class BulkTeamMemberBudgetUpdateRequest(BaseModel):
+class BulkTeamMemberBudgetUpdateRequest(LiteLLMBaseModel):
     """Body of `POST /management/v1/teams/{team_id}/members/bulk_update`."""
 
     model_config = ConfigDict(extra="forbid")
@@ -186,7 +187,7 @@ class BulkTeamMemberBudgetUpdateRequest(BaseModel):
     members: tuple[TeamMemberBudgetPatch, ...] = Field(min_length=1, max_length=MAX_BULK_TEAM_MEMBER_BUDGET_UPDATES)
 
 
-class TeamMemberBudgetUpdateResult(BaseModel):
+class TeamMemberBudgetUpdateResult(LiteLLMBaseModel):
     """Outcome for one requested member, in request order, carrying the limits in force
     after the write rather than the ones that were asked for."""
 
@@ -215,7 +216,7 @@ class TeamMemberInfoResponse(LiteLLM_TeamMembership):
     team_alias: str | None = None
 
 
-class TeamMetadataFieldSchema(BaseModel):
+class TeamMetadataFieldSchema(LiteLLMBaseModel):
     """One declared team metadata field from ``general_settings.team_metadata_schema``.
 
     Advisory only: the UI uses it to prepopulate the team metadata form.
@@ -228,13 +229,13 @@ class TeamMetadataFieldSchema(BaseModel):
     label: str | None = None
 
 
-class TeamMetadataSchemaResponse(BaseModel):
+class TeamMetadataSchemaResponse(LiteLLMBaseModel):
     """Response for GET /team/metadata_schema; ``fields`` is empty when no schema is configured."""
 
     fields: tuple[TeamMetadataFieldSchema, ...]
 
 
-class TeamUserSpendRow(BaseModel):
+class TeamUserSpendRow(LiteLLMBaseModel):
     team_id: str
     team_alias: str | None = None
     user_id: str
@@ -249,7 +250,7 @@ class TeamUserSpendRow(BaseModel):
     failed_requests: int = 0
 
 
-class TeamUserSpendResponse(BaseModel):
+class TeamUserSpendResponse(LiteLLMBaseModel):
     start_date: str
     end_date: str
     results: tuple[TeamUserSpendRow, ...]
diff --git a/litellm/types/proxy/model_access_group_budget.py b/litellm/types/proxy/model_access_group_budget.py
index cccbe92b5d6..08ea04f44f3 100644
--- a/litellm/types/proxy/model_access_group_budget.py
+++ b/litellm/types/proxy/model_access_group_budget.py
@@ -2,10 +2,10 @@
 
 from __future__ import annotations
 
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class ModelAccessGroupBudget(BaseModel):
+class ModelAccessGroupBudget(LiteLLMBaseModel):
     """One model access group's budget, flattened out of its joined ``LiteLLM_ModelAccessGroupBudgetTable`` row.
 
     Both readers want only the recorded spend and the ceiling, and this sits on the per-request hot
diff --git a/litellm/types/proxy/model_deprecation.py b/litellm/types/proxy/model_deprecation.py
index bbad63a278d..0b0cd2d7024 100644
--- a/litellm/types/proxy/model_deprecation.py
+++ b/litellm/types/proxy/model_deprecation.py
@@ -3,7 +3,9 @@ from __future__ import annotations
 from datetime import date, datetime
 from typing import Final, Literal
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 DEFAULT_DEPRECATION_WARN_DAYS: Final = 30
 
@@ -14,7 +16,7 @@ DEPRECATION_IDLE_POLL_SECONDS: Final = 30
 DeprecationStatus = Literal["upcoming", "imminent", "deprecated"]
 
 
-class ModelDeprecationInfo(BaseModel):
+class ModelDeprecationInfo(LiteLLMBaseModel):
     model_name: str = Field(description="The public name of the model on the proxy (model_group).")
     litellm_model: str | None = Field(
         default=None,
@@ -32,7 +34,7 @@ class ModelDeprecationInfo(BaseModel):
     litellm_provider: str | None = Field(default=None, description="The provider this model belongs to.")
 
 
-class ModelDeprecationResponse(BaseModel):
+class ModelDeprecationResponse(LiteLLMBaseModel):
     deprecated: list[ModelDeprecationInfo] = Field(
         default_factory=list,
         description="Models whose deprecation date has already passed.",
diff --git a/litellm/types/proxy/policy_engine/pipeline_types.py b/litellm/types/proxy/policy_engine/pipeline_types.py
index 063edb0a264..5278754701d 100644
--- a/litellm/types/proxy/policy_engine/pipeline_types.py
+++ b/litellm/types/proxy/policy_engine/pipeline_types.py
@@ -8,13 +8,15 @@ with configurable actions on pass/fail, rather than independently.
 
 from typing import Any, Final, Literal
 
-from pydantic import BaseModel, ConfigDict, Field, field_validator
+from pydantic import ConfigDict, Field, field_validator
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 VALID_PIPELINE_ACTIONS: Final = {"allow", "block", "next", "modify_response"}
 VALID_PIPELINE_MODES: Final = {"pre_call", "post_call"}
 
 
-class PipelineStep(BaseModel):
+class PipelineStep(LiteLLMBaseModel):
     """
     A single step in a guardrail pipeline.
 
@@ -57,7 +59,7 @@ class PipelineStep(BaseModel):
         return v
 
 
-class GuardrailPipeline(BaseModel):
+class GuardrailPipeline(LiteLLMBaseModel):
     """
     Defines ordered execution of guardrails with conditional actions.
 
@@ -81,7 +83,7 @@ class GuardrailPipeline(BaseModel):
         return v
 
 
-class PipelineStepResult(BaseModel):
+class PipelineStepResult(LiteLLMBaseModel):
     """Result of executing a single pipeline step."""
 
     guardrail_name: str
@@ -92,7 +94,7 @@ class PipelineStepResult(BaseModel):
     duration_seconds: float | None = None
 
 
-class PipelineExecutionResult(BaseModel):
+class PipelineExecutionResult(LiteLLMBaseModel):
     """Result of executing an entire pipeline."""
 
     model_config = ConfigDict(arbitrary_types_allowed=True)
diff --git a/litellm/types/proxy/policy_engine/policy_types.py b/litellm/types/proxy/policy_engine/policy_types.py
index 73eeffa3585..5e806556a4f 100644
--- a/litellm/types/proxy/policy_engine/policy_types.py
+++ b/litellm/types/proxy/policy_engine/policy_types.py
@@ -29,8 +29,9 @@ Key concepts:
 - `condition`: Optional model condition for when guardrails apply
 """
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
 
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.proxy.policy_engine.pipeline_types import GuardrailPipeline
 
 # ─────────────────────────────────────────────────────────────────────────────
@@ -38,7 +39,7 @@ from litellm.types.proxy.policy_engine.pipeline_types import GuardrailPipeline
 # ─────────────────────────────────────────────────────────────────────────────
 
 
-class PolicyCondition(BaseModel):
+class PolicyCondition(LiteLLMBaseModel):
     """
     Condition for when a policy's guardrails apply.
 
@@ -66,7 +67,7 @@ class PolicyCondition(BaseModel):
 # ─────────────────────────────────────────────────────────────────────────────
 
 
-class PolicyScope(BaseModel):
+class PolicyScope(LiteLLMBaseModel):
     """
     Defines the scope for matching requests.
 
@@ -130,7 +131,7 @@ class PolicyScope(BaseModel):
 # ─────────────────────────────────────────────────────────────────────────────
 
 
-class PolicyGuardrails(BaseModel):
+class PolicyGuardrails(LiteLLMBaseModel):
     """
     Defines guardrails to add or remove in a policy.
 
@@ -167,7 +168,7 @@ class PolicyGuardrails(BaseModel):
 # ─────────────────────────────────────────────────────────────────────────────
 
 
-class Policy(BaseModel):
+class Policy(LiteLLMBaseModel):
     """
     A policy that defines WHAT guardrails to apply.
 
@@ -244,7 +245,7 @@ class Policy(BaseModel):
 # ─────────────────────────────────────────────────────────────────────────────
 
 
-class PolicyAttachment(BaseModel):
+class PolicyAttachment(LiteLLMBaseModel):
     """
     Attaches a policy to a scope - defines WHERE a policy applies.
 
@@ -317,7 +318,7 @@ class PolicyAttachment(BaseModel):
         )
 
 
-class PolicyConfig(BaseModel):
+class PolicyConfig(LiteLLMBaseModel):
     """
     Root configuration for all policies.
 
diff --git a/litellm/types/proxy/policy_engine/resolver_types.py b/litellm/types/proxy/policy_engine/resolver_types.py
index ebdedb98b12..248e669e83a 100644
--- a/litellm/types/proxy/policy_engine/resolver_types.py
+++ b/litellm/types/proxy/policy_engine/resolver_types.py
@@ -8,10 +8,12 @@ the final guardrails list.
 from datetime import datetime
 from typing import Any, Literal
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class PolicyMatchContext(BaseModel):
+class PolicyMatchContext(LiteLLMBaseModel):
     """
     Context used to match a request against policies.
 
@@ -38,7 +40,7 @@ class PolicyMatchContext(BaseModel):
     model_config = ConfigDict(extra="forbid")
 
 
-class ResolvedPolicy(BaseModel):
+class ResolvedPolicy(LiteLLMBaseModel):
     """
     Result of resolving a policy with its inheritance chain.
 
@@ -63,7 +65,7 @@ class ResolvedPolicy(BaseModel):
 # ─────────────────────────────────────────────────────────────────────────────
 
 
-class PolicyScopeResponse(BaseModel):
+class PolicyScopeResponse(LiteLLMBaseModel):
     """Scope configuration for a policy."""
 
     teams: list[str] = Field(default_factory=list)
@@ -72,14 +74,14 @@ class PolicyScopeResponse(BaseModel):
     tags: list[str] = Field(default_factory=list)
 
 
-class PolicyGuardrailsResponse(BaseModel):
+class PolicyGuardrailsResponse(LiteLLMBaseModel):
     """Guardrails configuration for a policy."""
 
     add: list[str] = Field(default_factory=list)
     remove: list[str] = Field(default_factory=list)
 
 
-class PolicyInfoResponse(BaseModel):
+class PolicyInfoResponse(LiteLLMBaseModel):
     """Response for /policy/info/{policy_name} endpoint."""
 
     policy_name: str
@@ -90,7 +92,7 @@ class PolicyInfoResponse(BaseModel):
     inheritance_chain: list[str]
 
 
-class PolicySummaryItem(BaseModel):
+class PolicySummaryItem(LiteLLMBaseModel):
     """Summary of a single policy for list endpoint."""
 
     inherit: str | None = None
@@ -100,14 +102,14 @@ class PolicySummaryItem(BaseModel):
     inheritance_chain: list[str]
 
 
-class PolicyListResponse(BaseModel):
+class PolicyListResponse(LiteLLMBaseModel):
     """Response for /policy/list endpoint."""
 
     policies: dict[str, PolicySummaryItem]
     total_count: int
 
 
-class PolicyTestResponse(BaseModel):
+class PolicyTestResponse(LiteLLMBaseModel):
     """Response for /policy/test endpoint."""
 
     context: PolicyMatchContext
@@ -121,7 +123,7 @@ class PolicyTestResponse(BaseModel):
 # ─────────────────────────────────────────────────────────────────────────────
 
 
-class PolicyConditionRequest(BaseModel):
+class PolicyConditionRequest(LiteLLMBaseModel):
     """Condition for when a policy applies."""
 
     model: str | None = Field(
@@ -130,7 +132,7 @@ class PolicyConditionRequest(BaseModel):
     )
 
 
-class PolicyCreateRequest(BaseModel):
+class PolicyCreateRequest(LiteLLMBaseModel):
     """Request body for creating a new policy."""
 
     policy_name: str = Field(description="Unique name for the policy.")
@@ -160,7 +162,7 @@ class PolicyCreateRequest(BaseModel):
     )
 
 
-class PolicyUpdateRequest(BaseModel):
+class PolicyUpdateRequest(LiteLLMBaseModel):
     """Request body for updating a policy."""
 
     policy_name: str | None = Field(
@@ -193,7 +195,7 @@ class PolicyUpdateRequest(BaseModel):
     )
 
 
-class PolicyDBResponse(BaseModel):
+class PolicyDBResponse(LiteLLMBaseModel):
     """Response for a policy from the database."""
 
     policy_id: str = Field(description="Unique ID of the policy.")
@@ -226,7 +228,7 @@ class PolicyDBResponse(BaseModel):
     )
 
 
-class PolicyListDBResponse(BaseModel):
+class PolicyListDBResponse(LiteLLMBaseModel):
     """Response for listing policies from the database."""
 
     policies: list[PolicyDBResponse] = Field(default_factory=list, description="List of policies.")
@@ -238,7 +240,7 @@ class PolicyListDBResponse(BaseModel):
 # ─────────────────────────────────────────────────────────────────────────────
 
 
-class PolicyVersionCreateRequest(BaseModel):
+class PolicyVersionCreateRequest(LiteLLMBaseModel):
     """Request body for creating a new policy version (draft)."""
 
     source_policy_id: str | None = Field(
@@ -247,7 +249,7 @@ class PolicyVersionCreateRequest(BaseModel):
     )
 
 
-class PolicyVersionStatusUpdateRequest(BaseModel):
+class PolicyVersionStatusUpdateRequest(LiteLLMBaseModel):
     """Request body for updating a policy version's status."""
 
     version_status: str = Field(
@@ -255,7 +257,7 @@ class PolicyVersionStatusUpdateRequest(BaseModel):
     )
 
 
-class PolicyVersionListResponse(BaseModel):
+class PolicyVersionListResponse(LiteLLMBaseModel):
     """Response for listing all versions of a policy."""
 
     policy_name: str = Field(description="Name of the policy.")
@@ -265,7 +267,7 @@ class PolicyVersionListResponse(BaseModel):
     total_count: int = Field(default=0, description="Total number of versions.")
 
 
-class PolicyVersionCompareResponse(BaseModel):
+class PolicyVersionCompareResponse(LiteLLMBaseModel):
     """Response for comparing two policy versions."""
 
     version_a: PolicyDBResponse = Field(description="First version.")
@@ -281,7 +283,7 @@ class PolicyVersionCompareResponse(BaseModel):
 # ─────────────────────────────────────────────────────────────────────────────
 
 
-class PolicyAttachmentCreateRequest(BaseModel):
+class PolicyAttachmentCreateRequest(LiteLLMBaseModel):
     """Request body for creating a policy attachment."""
 
     policy_name: str = Field(description="Name of the policy to attach.")
@@ -317,7 +319,7 @@ class PolicyAttachmentCreateRequest(BaseModel):
     )
 
 
-class PolicyAttachmentDBResponse(BaseModel):
+class PolicyAttachmentDBResponse(LiteLLMBaseModel):
     """Response for a policy attachment from the database."""
 
     attachment_id: str = Field(description="Unique ID of the attachment.")
@@ -345,7 +347,7 @@ class PolicyAttachmentDBResponse(BaseModel):
     )
 
 
-class PolicyAttachmentListResponse(BaseModel):
+class PolicyAttachmentListResponse(LiteLLMBaseModel):
     """Response for listing policy attachments."""
 
     attachments: list[PolicyAttachmentDBResponse] = Field(
@@ -359,7 +361,7 @@ class PolicyAttachmentListResponse(BaseModel):
 # ─────────────────────────────────────────────────────────────────────────────
 
 
-class PipelineTestRequest(BaseModel):
+class PipelineTestRequest(LiteLLMBaseModel):
     """Request body for testing a guardrail pipeline with sample messages."""
 
     pipeline: dict[str, Any] = Field(
@@ -370,7 +372,7 @@ class PipelineTestRequest(BaseModel):
     )
 
 
-class PolicyResolveRequest(BaseModel):
+class PolicyResolveRequest(LiteLLMBaseModel):
     """Request body for resolving effective policies/guardrails for a context."""
 
     team_alias: str | None = Field(default=None, description="Team alias to resolve for.")
@@ -379,7 +381,7 @@ class PolicyResolveRequest(BaseModel):
     tags: list[str] | None = Field(default=None, description="Tags to resolve for.")
 
 
-class PolicyMatchDetail(BaseModel):
+class PolicyMatchDetail(LiteLLMBaseModel):
     """Details about why a specific policy matched."""
 
     policy_name: str = Field(description="Name of the matched policy.")
@@ -392,7 +394,7 @@ class PolicyMatchDetail(BaseModel):
     )
 
 
-class PolicyResolveResponse(BaseModel):
+class PolicyResolveResponse(LiteLLMBaseModel):
     """Response for resolving effective policies/guardrails for a context."""
 
     effective_guardrails: list[str] = Field(
@@ -410,7 +412,7 @@ class PolicyResolveResponse(BaseModel):
 # ─────────────────────────────────────────────────────────────────────────────
 
 
-class AttachmentImpactResponse(BaseModel):
+class AttachmentImpactResponse(LiteLLMBaseModel):
     """Response for estimating the impact of a policy attachment."""
 
     affected_keys_count: int = Field(
diff --git a/litellm/types/proxy/policy_engine/validation_types.py b/litellm/types/proxy/policy_engine/validation_types.py
index 1e4925e1d08..d9537c34084 100644
--- a/litellm/types/proxy/policy_engine/validation_types.py
+++ b/litellm/types/proxy/policy_engine/validation_types.py
@@ -8,7 +8,9 @@ validation results.
 from enum import Enum
 from typing import Any
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
 class PolicyValidationErrorType(str, Enum):
@@ -24,7 +26,7 @@ class PolicyValidationErrorType(str, Enum):
     INVALID_SYNTAX = "invalid_syntax"
 
 
-class PolicyValidationError(BaseModel):
+class PolicyValidationError(LiteLLMBaseModel):
     """
     Represents a validation error or warning for a policy.
     """
@@ -44,7 +46,7 @@ class PolicyValidationError(BaseModel):
     model_config = ConfigDict(extra="forbid")
 
 
-class PolicyValidationResponse(BaseModel):
+class PolicyValidationResponse(LiteLLMBaseModel):
     """
     Response from policy validation.
 
@@ -66,7 +68,7 @@ class PolicyValidationResponse(BaseModel):
     model_config = ConfigDict(extra="forbid")
 
 
-class PolicyValidateRequest(BaseModel):
+class PolicyValidateRequest(LiteLLMBaseModel):
     """
     Request body for the /policy/validate endpoint.
     """
diff --git a/litellm/types/proxy/prompt_endpoints.py b/litellm/types/proxy/prompt_endpoints.py
index ba2c5d3c373..930713663a1 100644
--- a/litellm/types/proxy/prompt_endpoints.py
+++ b/litellm/types/proxy/prompt_endpoints.py
@@ -1,9 +1,9 @@
 from typing import Any
 
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class TestPromptRequest(BaseModel):
+class TestPromptRequest(LiteLLMBaseModel):
     dotprompt_content: str
     prompt_variables: dict[str, Any] | None = None
     conversation_history: list[dict[str, str]] | None = None
diff --git a/litellm/types/proxy/public_endpoints/public_endpoints.py b/litellm/types/proxy/public_endpoints/public_endpoints.py
index fa73926305b..c487ce8b6c5 100644
--- a/litellm/types/proxy/public_endpoints/public_endpoints.py
+++ b/litellm/types/proxy/public_endpoints/public_endpoints.py
@@ -1,10 +1,12 @@
 from collections.abc import Mapping, Sequence
 from typing import Any, Literal
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class PublicModelHubInfo(BaseModel):
+class PublicModelHubInfo(LiteLLMBaseModel):
     docs_title: str
     custom_docs_description: str | None
     litellm_version: str
@@ -14,7 +16,7 @@ class PublicModelHubInfo(BaseModel):
     useful_links: dict[str, str | dict[str, Any]] | None
 
 
-class ProviderCredentialField(BaseModel):
+class ProviderCredentialField(LiteLLMBaseModel):
     key: str
     label: str
     placeholder: str | None = None
@@ -25,7 +27,7 @@ class ProviderCredentialField(BaseModel):
     default_value: str | None = None
 
 
-class ProviderCreateInfo(BaseModel):
+class ProviderCreateInfo(LiteLLMBaseModel):
     provider: str
     provider_display_name: str
     litellm_provider: str
@@ -33,7 +35,7 @@ class ProviderCreateInfo(BaseModel):
     default_model_placeholder: str | None = None
 
 
-class AgentCredentialField(BaseModel):
+class AgentCredentialField(LiteLLMBaseModel):
     key: str
     label: str
     placeholder: str | None = None
@@ -47,7 +49,7 @@ class AgentCredentialField(BaseModel):
     validation_message: str | None = None
 
 
-class AgentCreateInfo(BaseModel):
+class AgentCreateInfo(LiteLLMBaseModel):
     agent_type: str
     agent_type_display_name: str
     description: str | None = None
@@ -57,23 +59,23 @@ class AgentCreateInfo(BaseModel):
     model_template: str | None = None
 
 
-class EndpointProvider(BaseModel):
+class EndpointProvider(LiteLLMBaseModel):
     slug: str
     display_name: str
 
 
-class SupportedEndpoint(BaseModel):
+class SupportedEndpoint(LiteLLMBaseModel):
     key: str
     label: str
     endpoint: str
     providers: list[EndpointProvider]
 
 
-class SupportedEndpointsResponse(BaseModel):
+class SupportedEndpointsResponse(LiteLLMBaseModel):
     endpoints: list[SupportedEndpoint]
 
 
-class AutoRouterPresetTiers(BaseModel):
+class AutoRouterPresetTiers(LiteLLMBaseModel):
     """Exactly the four built-in tiers the dashboard's preset prefill can apply.
 
     extra="forbid" on purpose: a tier name this dashboard cannot apply would grey out or crash the
@@ -88,7 +90,7 @@ class AutoRouterPresetTiers(BaseModel):
     REASONING: Sequence[str]
 
 
-class AutoRouterPresetConfig(BaseModel):
+class AutoRouterPresetConfig(LiteLLMBaseModel):
     """The complexity_router_config a preset prefills.
 
     Only tiers is validated, because every dashboard consumer dereferences it; everything else
@@ -101,7 +103,7 @@ class AutoRouterPresetConfig(BaseModel):
     tiers: AutoRouterPresetTiers
 
 
-class AutoRouterPresetRecord(BaseModel):
+class AutoRouterPresetRecord(LiteLLMBaseModel):
     """One auto-router preset as served to the dashboard's template picker."""
 
     model_config = ConfigDict(extra="allow")
@@ -111,7 +113,7 @@ class AutoRouterPresetRecord(BaseModel):
     complexity_router_config: AutoRouterPresetConfig
 
 
-class ComplexityScorerDefaults(BaseModel):
+class ComplexityScorerDefaults(LiteLLMBaseModel):
     """The complexity router's shipped heuristic scorer defaults.
 
     The dashboard prefills its Advanced scoring controls from these rather than keeping its own copy, so
diff --git a/litellm/types/proxy/spend_capture_rate.py b/litellm/types/proxy/spend_capture_rate.py
index ccc9daf8a9f..54fa6daf8d1 100644
--- a/litellm/types/proxy/spend_capture_rate.py
+++ b/litellm/types/proxy/spend_capture_rate.py
@@ -7,14 +7,15 @@ higher means LiteLLM prices above the bill. ``None`` means the provider billed n
 
 from typing import Literal
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
 
 from litellm.constants import SPEND_CAPTURE_RATE_MAX_RANGE_DAYS
+from litellm.types.llms.base import LiteLLMBaseModel
 
 SpendCaptureProvider = Literal["openai"]
 
 
-class SpendCaptureRateCheckSettings(BaseModel):
+class SpendCaptureRateCheckSettings(LiteLLMBaseModel):
     """``general_settings.spend_capture_rate_check``: the daily check of captured spend against the provider bill."""
 
     model_config = ConfigDict(frozen=True, extra="forbid")
@@ -31,7 +32,7 @@ class SpendCaptureRateCheckSettings(BaseModel):
     )
 
 
-class CaptureRateDay(BaseModel):
+class CaptureRateDay(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     date: str
@@ -40,7 +41,7 @@ class CaptureRateDay(BaseModel):
     capture_rate: float | None
 
 
-class CaptureRateReport(BaseModel):
+class CaptureRateReport(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     provider: SpendCaptureProvider
diff --git a/litellm/types/proxy/vantage_endpoints.py b/litellm/types/proxy/vantage_endpoints.py
index 83bd5c4c61b..1833d8b8386 100644
--- a/litellm/types/proxy/vantage_endpoints.py
+++ b/litellm/types/proxy/vantage_endpoints.py
@@ -5,10 +5,12 @@ Vantage endpoint types for LiteLLM Proxy
 from datetime import datetime
 from typing import Any
 
-from pydantic import BaseModel, Field, field_validator
+from pydantic import Field, field_validator
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class VantageInitRequest(BaseModel):
+class VantageInitRequest(LiteLLMBaseModel):
     """Request model for initializing Vantage settings"""
 
     api_key: str = Field(..., description="Vantage API key for authentication")
@@ -26,14 +28,14 @@ class VantageInitRequest(BaseModel):
         return v
 
 
-class VantageInitResponse(BaseModel):
+class VantageInitResponse(LiteLLMBaseModel):
     """Response model for Vantage initialization"""
 
     message: str
     status: str
 
 
-class VantageExportRequest(BaseModel):
+class VantageExportRequest(LiteLLMBaseModel):
     """Request model for Vantage export operations (actual export, no default limit)"""
 
     limit: int | None = Field(
@@ -44,13 +46,13 @@ class VantageExportRequest(BaseModel):
     end_time_utc: datetime | None = Field(None, description="End time for data export in UTC")
 
 
-class VantageDryRunRequest(BaseModel):
+class VantageDryRunRequest(LiteLLMBaseModel):
     """Request model for Vantage dry-run operations (capped for preview)"""
 
     limit: int | None = Field(500, description="Limit on number of records to preview (default: 500)")
 
 
-class VantageExportResponse(BaseModel):
+class VantageExportResponse(LiteLLMBaseModel):
     """Response model for Vantage export operations"""
 
     message: str
@@ -61,7 +63,7 @@ class VantageExportResponse(BaseModel):
     summary: dict[str, Any] | None = Field(None, description="Summary statistics for dry run")
 
 
-class VantageSettingsView(BaseModel):
+class VantageSettingsView(LiteLLMBaseModel):
     """Response model for viewing Vantage settings with masked API key"""
 
     api_key_masked: str | None = Field(
@@ -76,7 +78,7 @@ class VantageSettingsView(BaseModel):
     status: str | None = Field(None, description="Configuration status")
 
 
-class VantageSettingsUpdate(BaseModel):
+class VantageSettingsUpdate(LiteLLMBaseModel):
     """Request model for updating Vantage settings"""
 
     api_key: str | None = Field(None, description="New Vantage API key for authentication")
diff --git a/litellm/types/rag.py b/litellm/types/rag.py
index 629979afde9..00c070b0996 100644
--- a/litellm/types/rag.py
+++ b/litellm/types/rag.py
@@ -5,9 +5,10 @@ Type definitions for RAG (Retrieval Augmented Generation) Ingest API.
 from collections.abc import Mapping
 from typing import Any, Literal
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import ConfigDict
 from typing_extensions import ReadOnly, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.utils import ModelResponse
 
 
@@ -225,7 +226,7 @@ class RAGIngestResponse(TypedDict, total=False):
     error: str | None  # Error message if status is "failed"
 
 
-class RAGIngestRequest(BaseModel):
+class RAGIngestRequest(LiteLLMBaseModel):
     """Request body for RAG ingest API (for validation)."""
 
     file_url: str | None = None  # URL to fetch file from
@@ -254,7 +255,7 @@ class RAGRerankConfig(TypedDict, total=False):
     return_documents: bool | None
 
 
-class RAGQueryRequest(BaseModel):
+class RAGQueryRequest(LiteLLMBaseModel):
     """Request body for RAG query API."""
 
     model: str
diff --git a/litellm/types/realtime.py b/litellm/types/realtime.py
index 855cccc8ddd..6da73982291 100644
--- a/litellm/types/realtime.py
+++ b/litellm/types/realtime.py
@@ -1,8 +1,10 @@
 from typing import Any, Literal
 
-from pydantic import BaseModel
+from pydantic import ConfigDict
 from typing_extensions import ReadOnly, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 from .llms.openai import (
     OpenAIRealtimeEvents,
     OpenAIRealtimeOutputItemDone,
@@ -57,14 +59,14 @@ class RealtimeQueryParams(TypedDict, total=False):
 # ---------------------------------------------------------------------------
 
 
-class RealtimeExpiresAfter(BaseModel):
+class RealtimeExpiresAfter(LiteLLMBaseModel):
     """Expiration config for a client secret."""
 
     anchor: str | None = "created_at"
     seconds: int | None = None
 
 
-class RealtimeSessionConfig(BaseModel):
+class RealtimeSessionConfig(LiteLLMBaseModel):
     """
     Session configuration nested inside the client_secrets request body.
 
@@ -73,7 +75,7 @@ class RealtimeSessionConfig(BaseModel):
     Extra/unknown fields are passed through unchanged.
     """
 
-    model_config = {"extra": "allow"}
+    model_config = ConfigDict(extra="allow")
 
     type: str | None = None
     model: str | None = None
@@ -89,7 +91,7 @@ class RealtimeSessionConfig(BaseModel):
     prompt: dict[str, object] | None = None
 
 
-class RealtimeClientSecretRequest(BaseModel):
+class RealtimeClientSecretRequest(LiteLLMBaseModel):
     """
     Request body for POST /v1/realtime/client_secrets.
 
@@ -103,7 +105,7 @@ class RealtimeClientSecretRequest(BaseModel):
     model: str | None = None
 
 
-class RealtimeClientSecretResponse(BaseModel):
+class RealtimeClientSecretResponse(LiteLLMBaseModel):
     """
     Response from POST /v1/realtime/client_secrets.
 
@@ -117,7 +119,7 @@ class RealtimeClientSecretResponse(BaseModel):
     session: dict[str, object] | None = None
 
 
-class RealtimeTranscriptionSessionRequest(BaseModel):
+class RealtimeTranscriptionSessionRequest(LiteLLMBaseModel):
     """
     Request body for POST /v1/realtime/transcription_sessions.
 
@@ -127,7 +129,7 @@ class RealtimeTranscriptionSessionRequest(BaseModel):
     unchanged to the provider.
     """
 
-    model_config = {"extra": "allow"}
+    model_config = ConfigDict(extra="allow")
 
     # LiteLLM-only routing hint — stripped before forwarding upstream.
     model: str | None = None
@@ -141,7 +143,7 @@ class RealtimeTranscriptionSessionRequest(BaseModel):
         return None
 
 
-class RealtimeTranscriptionSessionResponse(BaseModel):
+class RealtimeTranscriptionSessionResponse(LiteLLMBaseModel):
     """
     Response from POST /v1/realtime/transcription_sessions.
 
@@ -149,7 +151,7 @@ class RealtimeTranscriptionSessionResponse(BaseModel):
     ephemeral key. Unknown fields pass through unchanged.
     """
 
-    model_config = {"extra": "allow"}
+    model_config = ConfigDict(extra="allow")
 
     client_secret: dict[str, object] | None = None
 
diff --git a/litellm/types/rerank.py b/litellm/types/rerank.py
index a76e6cf1187..81cb1b7e315 100644
--- a/litellm/types/rerank.py
+++ b/litellm/types/rerank.py
@@ -6,11 +6,13 @@ https://docs.cohere.com/reference/rerank
 
 from typing import Literal
 
-from pydantic import BaseModel, ConfigDict, PrivateAttr
+from pydantic import ConfigDict, PrivateAttr
 from typing_extensions import ReadOnly, Required, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
 
-class RerankRequest(BaseModel):
+
+class RerankRequest(LiteLLMBaseModel):
     model: str
     query: str
     top_n: int | None = None
@@ -28,7 +30,7 @@ class RerankRequest(BaseModel):
     max_tokens_per_query: int | None = None
 
 
-class HostedVLLMRerankTruncationParams(BaseModel):
+class HostedVLLMRerankTruncationParams(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     truncate_prompt_tokens: int | None = None
@@ -77,7 +79,7 @@ class RerankResponseResult(TypedDict, total=False):
     document: RerankResponseDocument
 
 
-class RerankResponse(BaseModel):
+class RerankResponse(LiteLLMBaseModel):
     id: str | None = None
     results: list[RerankResponseResult] | None = None  # Contains index and relevance_score
     meta: RerankResponseMeta | None = None  # Contains api_version and billed_units
diff --git a/litellm/types/roi_calculator.py b/litellm/types/roi_calculator.py
index 7a6cd8ab2e1..d8c0f27a895 100644
--- a/litellm/types/roi_calculator.py
+++ b/litellm/types/roi_calculator.py
@@ -3,9 +3,11 @@ from datetime import datetime
 from types import MappingProxyType
 from typing import Final, Literal
 
-from pydantic import BaseModel, ConfigDict, Field, SecretStr, StrictFloat, StrictInt, ValidationInfo, field_validator
+from pydantic import ConfigDict, Field, SecretStr, StrictFloat, StrictInt, ValidationInfo, field_validator
 from typing_extensions import NotRequired, ReadOnly, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 DEFAULT_PROMPT: Final = (
     "Estimate how many hours it would take an engineer to complete the work in this pull request without AI assistance. "
     "Explain your estimate briefly."
@@ -22,7 +24,7 @@ def normalize_source_login(value: str, provider: str = "github") -> str:
     return login
 
 
-class ROISettings(BaseModel):
+class ROISettings(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
     report_mode: Literal["legacy", "observed"] = "legacy"
@@ -129,7 +131,7 @@ class ROISettings(BaseModel):
         return normalized
 
 
-class ROISettingsUpdate(BaseModel):
+class ROISettingsUpdate(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid")
 
     report_mode: Literal["legacy", "observed"] | None = None
@@ -146,12 +148,12 @@ class ROISettingsUpdate(BaseModel):
     update_interval_minutes: float | None = Field(default=None, ge=0, le=43200, allow_inf_nan=False)
 
 
-class ROIEstimatorModel(BaseModel):
+class ROIEstimatorModel(LiteLLMBaseModel):
     model_name: str
     provider_models: tuple[str, ...]
 
 
-class ROISettingsResponse(BaseModel):
+class ROISettingsResponse(LiteLLMBaseModel):
     report_mode: Literal["legacy", "observed"] = "legacy"
     source_provider: Literal["github", "gitlab"] = "github"
     gitlab_api_url: str = "https://gitlab.com/api/v4"
@@ -171,19 +173,19 @@ class ROISettingsResponse(BaseModel):
     ready: bool
 
 
-class ROIRepository(BaseModel):
+class ROIRepository(LiteLLMBaseModel):
     name: str
     visibility: str
     archived: bool
 
 
-class ROIRepositoriesResponse(BaseModel):
+class ROIRepositoriesResponse(LiteLLMBaseModel):
     repositories: tuple[ROIRepository, ...]
     page: int
     has_more: bool
 
 
-class ROISyncStatus(BaseModel):
+class ROISyncStatus(LiteLLMBaseModel):
     running: bool
     phase: Literal["idle", "spend", "repositories", "estimates", "complete", "cancelled", "error"]
     stage: str
@@ -240,14 +242,14 @@ class ROIPullRecord(TypedDict):
     cache_key: ReadOnly[str | None]
 
 
-class ROIBranchSpend(BaseModel):
+class ROIBranchSpend(LiteLLMBaseModel):
     repo: str
     branch: str
     spend: float
     requests: int
 
 
-class ROIBranchAttribution(BaseModel):
+class ROIBranchAttribution(LiteLLMBaseModel):
     repo: str
     branch: str
     spend: float | None = None
@@ -255,7 +257,7 @@ class ROIBranchAttribution(BaseModel):
     status: Literal["matched", "unattributed", "ambiguous", "unavailable"] = "unattributed"
 
 
-class ROIBranchMetrics(BaseModel):
+class ROIBranchMetrics(LiteLLMBaseModel):
     spend: float = 0
     hours: float = 0
     cost_per_hour: float | None = None
@@ -410,7 +412,7 @@ class ROISummary(TypedDict):
     trend: ReadOnly[tuple[ROITrendDay, ...]]
 
 
-class ROIMetricsResponse(BaseModel):
+class ROIMetricsResponse(LiteLLMBaseModel):
     matched_spend: float
     output_hours: float
     total_spend: float
@@ -426,7 +428,7 @@ class ROIMetricsResponse(BaseModel):
     pending_prs: int
 
 
-class ROIPersonResponse(BaseModel):
+class ROIPersonResponse(LiteLLMBaseModel):
     id: str
     email: str
     logins: tuple[str, ...]
@@ -440,7 +442,7 @@ class ROIPersonResponse(BaseModel):
     cost_per_hour: float | None
 
 
-class ROIEstimateResponse(BaseModel):
+class ROIEstimateResponse(LiteLLMBaseModel):
     status: Literal["estimated", "needs_review", "error"]
     hours: float | None
     reasoning: str
@@ -450,7 +452,7 @@ class ROIEstimateResponse(BaseModel):
     cached: bool = False
 
 
-class ROIPullResponse(BaseModel):
+class ROIPullResponse(LiteLLMBaseModel):
     source_repo: str = ""
     source_branch: str = ""
     branch_cost: ROIBranchAttribution = Field(default_factory=lambda: ROIBranchAttribution(repo="", branch=""))
@@ -475,14 +477,14 @@ class ROIPullResponse(BaseModel):
     matched: bool
 
 
-class ROITrendResponse(BaseModel):
+class ROITrendResponse(LiteLLMBaseModel):
     date: str
     spend: float
     hours: float
     prs: int
 
 
-class ROISummaryResponse(BaseModel):
+class ROISummaryResponse(LiteLLMBaseModel):
     source_provider: Literal["github", "gitlab"] = "github"
     branch_metrics: ROIBranchMetrics = Field(default_factory=ROIBranchMetrics)
     unlinked_branches: tuple[ROIBranchSpend, ...] = ()
@@ -502,11 +504,11 @@ class ROISummaryResponse(BaseModel):
     trend: tuple[ROITrendResponse, ...]
 
 
-class ROIReportResponse(BaseModel):
+class ROIReportResponse(LiteLLMBaseModel):
     report: ROISummaryResponse | None
 
 
-class ROIIdentityMapUpdate(BaseModel):
+class ROIIdentityMapUpdate(LiteLLMBaseModel):
     github_login: str
     email: str | None
 
@@ -516,26 +518,26 @@ class ROIIdentityMapUpdate(BaseModel):
         return value.strip().casefold()
 
 
-class ROIIdentityMapResponse(BaseModel):
+class ROIIdentityMapResponse(LiteLLMBaseModel):
     report: ROISummaryResponse | None
     identity_map: Mapping[str, str]
 
 
-class ROIEstimatorChanges(BaseModel):
+class ROIEstimatorChanges(LiteLLMBaseModel):
     additions: int
     deletions: int
     files: int
     commits: int
 
 
-class ROIEstimatorFile(BaseModel):
+class ROIEstimatorFile(LiteLLMBaseModel):
     filename: str | None
     status: str | None
     additions: int | None
     deletions: int | None
 
 
-class ROIEstimatorCommit(BaseModel):
+class ROIEstimatorCommit(LiteLLMBaseModel):
     sha: str
     message: str
     additions: int | None = None
@@ -543,7 +545,7 @@ class ROIEstimatorCommit(BaseModel):
     changed_files: int | None = None
 
 
-class ROIEstimatorEvidence(BaseModel):
+class ROIEstimatorEvidence(LiteLLMBaseModel):
     repo: str
     number: int
     title: str
@@ -566,7 +568,7 @@ class ROIResponseFormat(TypedDict):
     type: ReadOnly[Literal["json_object"]]
 
 
-class ROICompletionRequest(BaseModel):
+class ROICompletionRequest(LiteLLMBaseModel):
     model: str
     temperature: Literal[0]
     messages: tuple[ROICompletionMessage, ...]
@@ -576,26 +578,26 @@ class ROICompletionRequest(BaseModel):
     reasoning_effort: Literal["none"] | None = None
 
 
-class _ROICompletionMessageResponse(BaseModel):
+class _ROICompletionMessageResponse(LiteLLMBaseModel):
     model_config = ConfigDict(from_attributes=True)
 
     content: str | None = None
 
 
-class _ROICompletionChoice(BaseModel):
+class _ROICompletionChoice(LiteLLMBaseModel):
     model_config = ConfigDict(from_attributes=True)
 
     finish_reason: str | None = None
     message: _ROICompletionMessageResponse
 
 
-class ROICompletionResponse(BaseModel):
+class ROICompletionResponse(LiteLLMBaseModel):
     model_config = ConfigDict(from_attributes=True)
 
     choices: tuple[_ROICompletionChoice, ...]
 
 
-class ROIEstimatorResult(BaseModel):
+class ROIEstimatorResult(LiteLLMBaseModel):
     model_config = ConfigDict(strict=True, extra="forbid")
 
     hours: StrictInt | StrictFloat
diff --git a/litellm/types/roi_observed.py b/litellm/types/roi_observed.py
index b617175788b..eee85d831a2 100644
--- a/litellm/types/roi_observed.py
+++ b/litellm/types/roi_observed.py
@@ -2,12 +2,13 @@ from collections.abc import Mapping
 from datetime import date, datetime
 from typing import Literal
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import ConfigDict, Field
 
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.roi_calculator import ROIBranchAttribution, ROIBranchSpend, ROISpendRecord
 
 
-class ObservedModel(BaseModel):
+class ObservedModel(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True)
 
 
@@ -151,7 +152,7 @@ class ObservedReportResponse(ObservedModel):
     report: ObservedReport | None
 
 
-class ObservedIdentityUpdate(BaseModel):
+class ObservedIdentityUpdate(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid")
 
     email: str
@@ -186,7 +187,7 @@ class ObservedSettings(ObservedConnection):
     connections: tuple[ObservedConnection, ...] = ()
 
 
-class ObservedSettingsUpdate(BaseModel):
+class ObservedSettingsUpdate(LiteLLMBaseModel):
     model_config = ConfigDict(extra="forbid")
 
     connection_id: str | None = Field(default=None, max_length=100)
diff --git a/litellm/types/router.py b/litellm/types/router.py
index 6c52120c3ae..8fd53079156 100644
--- a/litellm/types/router.py
+++ b/litellm/types/router.py
@@ -21,13 +21,14 @@ from typing import (
 from zoneinfo import ZoneInfo, ZoneInfoNotFoundError
 
 import httpx
-from pydantic import BaseModel, ConfigDict, Field, field_validator, model_validator
+from pydantic import ConfigDict, Field, field_validator, model_validator
 from typing_extensions import Protocol, ReadOnly, Required, TypedDict, runtime_checkable
 
 from litellm._logging import verbose_logger
 from litellm._uuid import uuid
 from litellm.litellm_core_utils.core_helpers import normalize_drop_params
 from litellm.litellm_core_utils.provider_affinity import validate_provider_affinity_header_name
+from litellm.types.llms.base import LiteLLMBaseModel
 from litellm.types.router_weights import RouterWeights
 
 if TYPE_CHECKING:
@@ -60,7 +61,7 @@ class ConfigurableClientsideParamsCustomAuth(TypedDict):
 CONFIGURABLE_CLIENTSIDE_AUTH_PARAMS = list[str | ConfigurableClientsideParamsCustomAuth] | None
 
 
-class ModelConfig(BaseModel):
+class ModelConfig(LiteLLMBaseModel):
     model_name: str
     litellm_params: CompletionRequest | EmbeddingRequest
     tpm: int
@@ -69,7 +70,7 @@ class ModelConfig(BaseModel):
     model_config = ConfigDict(protected_namespaces=())
 
 
-class RoutingGroup(BaseModel):
+class RoutingGroup(LiteLLMBaseModel):
     """
     A group of models that share a routing strategy.
     """
@@ -101,7 +102,7 @@ class RoutingGroup(BaseModel):
     model_config = ConfigDict(protected_namespaces=())
 
 
-class RouterConfig(BaseModel):
+class RouterConfig(LiteLLMBaseModel):
     model_list: list[ModelConfig]
 
     redis_url: str | None = None
@@ -110,17 +111,17 @@ class RouterConfig(BaseModel):
     redis_password: str | None = None
 
     cache_responses: bool | None = False
-    cache_kwargs: dict | None = {}
+    cache_kwargs: dict | None = Field(default={})
     caching_groups: list[tuple[str, list[str]]] | None = None
     client_ttl: int | None = 3600
     num_retries: int | None = 0
     timeout: float | None = None
-    default_litellm_params: dict[str, str] | None = {}
+    default_litellm_params: dict[str, str] | None = Field(default={})
     set_verbose: bool | None = False
-    fallbacks: list | None = []
+    fallbacks: list | None = Field(default=[])
     allowed_fails: int | None = None
-    context_window_fallbacks: list | None = []
-    model_group_alias: dict[str, list[str]] | None = {}
+    context_window_fallbacks: list | None = Field(default=[])
+    model_group_alias: dict[str, list[str]] | None = Field(default={})
     retry_after: int | None = 0
     routing_strategy: RoutingStrategyName = "simple-shuffle"
     routing_groups: list[RoutingGroup] | None = None
@@ -128,7 +129,7 @@ class RouterConfig(BaseModel):
     model_config = ConfigDict(protected_namespaces=())
 
 
-class RetryPolicy(BaseModel):
+class RetryPolicy(LiteLLMBaseModel):
     """
     Use this to set a custom number of retries per exception type
     If RateLimitErrorRetries = 3, then 3 retries will be made for RateLimitError
@@ -161,7 +162,7 @@ OptionalPreCallChecks = list[
 ]
 
 
-class UpdateRouterConfig(BaseModel):
+class UpdateRouterConfig(LiteLLMBaseModel):
     """
     Set of params that you can modify via `router.update_settings()`.
     """
@@ -180,7 +181,7 @@ class UpdateRouterConfig(BaseModel):
     retry_after: float | None = None
     fallbacks: list[dict] | None = None
     context_window_fallbacks: list[dict] | None = None
-    model_group_alias: dict[str, str | dict] | None = {}
+    model_group_alias: dict[str, str | dict] | None = Field(default={})
     enable_tag_filtering: bool | None = None
     weights: RouterWeights | None = None
     tag_routing_prefix: str | None = None
@@ -197,7 +198,7 @@ def _as_utc(value: datetime.datetime | None) -> datetime.datetime | None:
     return value.astimezone(datetime.timezone.utc)
 
 
-class ModelAccessWindow(BaseModel):
+class ModelAccessWindow(LiteLLMBaseModel):
     model_config = ConfigDict(frozen=True, extra="forbid")
 
     start: datetime.time
@@ -336,7 +337,7 @@ class ModelInfo(MirroredPricingParams):
         setattr(self, key, value)
 
 
-class CredentialLiteLLMParams(BaseModel):
+class CredentialLiteLLMParams(LiteLLMBaseModel):
     api_key: str | None = None
     api_base: str | None = None
     api_version: str | None = None
@@ -630,7 +631,7 @@ class updateLiteLLMParams(GenericLiteLLMParams):
     model: str | None = None
 
 
-class updateDeployment(BaseModel):
+class updateDeployment(LiteLLMBaseModel):
     model_name: str | None = None
     litellm_params: updateLiteLLMParams | None = None
     model_info: ModelInfo | None = None
@@ -724,7 +725,7 @@ class DeploymentTypedDict(TypedDict, total=False):
 SPECIAL_MODEL_INFO_PARAMS = tuple(MirroredPricingParams.model_fields)
 
 
-class Deployment(BaseModel):
+class Deployment(LiteLLMBaseModel):
     model_name: str
     litellm_params: LiteLLM_Params
     model_info: ModelInfo
@@ -824,7 +825,7 @@ class RouterErrors(enum.Enum):
     )
 
 
-class AllowedFailsPolicy(BaseModel):
+class AllowedFailsPolicy(LiteLLMBaseModel):
     """
     Use this to set a custom number of allowed fails/minute before cooling down a deployment
     If `AuthenticationErrorAllowedFails = 1000`, then 1000 AuthenticationError will be allowed before cooling down a deployment
@@ -844,7 +845,7 @@ class AllowedFailsPolicy(BaseModel):
     NotFoundErrorAllowedFails: int | None = None
 
 
-class AlertingConfig(BaseModel):
+class AlertingConfig(LiteLLMBaseModel):
     """
     Use this configure alerting for the router. Receive alerts on the following events
     - LLM API Exceptions
@@ -865,7 +866,7 @@ def _resolved_annotations(model_class: type[object]) -> Mapping[str, object]:
     return get_type_hints(model_class)
 
 
-class ModelGroupInfo(BaseModel):
+class ModelGroupInfo(LiteLLMBaseModel):
     model_group: str
     providers: list[str]
     max_input_tokens: float | None = None
@@ -966,7 +967,7 @@ class GuardrailTypedDict(TypedDict, total=False):
     id: str | None  # Unique identifier for the guardrail deployment
 
 
-class FineTuningConfig(BaseModel):
+class FineTuningConfig(LiteLLMBaseModel):
     custom_llm_provider: Literal["azure", "openai"]
 
 
@@ -1018,7 +1019,7 @@ class CustomRoutingStrategyBase:
         """
 
 
-class RouterGeneralSettings(BaseModel):
+class RouterGeneralSettings(LiteLLMBaseModel):
     async_only_mode: bool = Field(default=False)  # this will only initialize async clients. Good for memory utils
     pass_through_all_models: bool = Field(
         default=False
@@ -1109,7 +1110,7 @@ class RouterCacheEnum(enum.Enum):
     OTPM = "global_router:{id}:{model}:otpm:{current_minute}"
 
 
-class GenericBudgetWindowDetails(BaseModel):
+class GenericBudgetWindowDetails(LiteLLMBaseModel):
     """Details about a provider's budget window"""
 
     budget_start: float
@@ -1184,11 +1185,11 @@ class MockRouterTestingParams:
         )
 
 
-class ModelGroupSettings(BaseModel):
+class ModelGroupSettings(LiteLLMBaseModel):
     forward_client_headers_to_llm_api: list[str] | None = None
 
 
-class PreRoutingHookResponse(BaseModel):
+class PreRoutingHookResponse(LiteLLMBaseModel):
     """
     Response object from the pre-routing hook.
 
@@ -1244,7 +1245,7 @@ class PreRoutingStrategy(Protocol):
     ) -> "PreRoutingHookResponse | None": ...
 
 
-class RoutingContext(BaseModel):
+class RoutingContext(LiteLLMBaseModel):
     """
     Passed through a Router's `plugins` pipeline before the routing decision is made.
 
@@ -1301,7 +1302,7 @@ class RequestType(str, enum.Enum):
     GENERAL = "general"
 
 
-class AdaptiveRouterWeights(BaseModel):
+class AdaptiveRouterWeights(LiteLLMBaseModel):
     quality: float = Field(default=0.7, ge=0.0, le=1.0)
     cost: float = Field(default=0.3, ge=0.0, le=1.0)
 
@@ -1314,12 +1315,12 @@ class AdaptiveRouterWeights(BaseModel):
         return v
 
 
-class AdaptiveRouterConfig(BaseModel):
+class AdaptiveRouterConfig(LiteLLMBaseModel):
     available_models: list[str]
     weights: AdaptiveRouterWeights = Field(default_factory=AdaptiveRouterWeights)
 
 
-class AdaptiveRouterPreferences(BaseModel):
+class AdaptiveRouterPreferences(LiteLLMBaseModel):
     """model_info.adaptive_router_preferences — declared by each model."""
 
     model_config = ConfigDict(use_enum_values=False)
diff --git a/litellm/types/services.py b/litellm/types/services.py
index 8a74be06da6..e86044c9303 100644
--- a/litellm/types/services.py
+++ b/litellm/types/services.py
@@ -1,9 +1,11 @@
 import enum
 from typing import Final
 
-from pydantic import BaseModel, Field
+from pydantic import Field
 from typing_extensions import TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 
 class ServiceMetrics(enum.Enum):
     COUNTER = "counter"
@@ -95,7 +97,7 @@ class ServiceEventMetadata(TypedDict, total=False):
     gauge_value: float | None
 
 
-class ServiceLoggerPayload(BaseModel):
+class ServiceLoggerPayload(LiteLLMBaseModel):
     """
     The payload logged during service success/failure
     """
diff --git a/litellm/types/tag_management.py b/litellm/types/tag_management.py
index f121f5bc562..e59b02af71a 100644
--- a/litellm/types/tag_management.py
+++ b/litellm/types/tag_management.py
@@ -1,7 +1,7 @@
-from pydantic import BaseModel
+from litellm.types.llms.base import LiteLLMBaseModel
 
 
-class TagBase(BaseModel):
+class TagBase(LiteLLMBaseModel):
     name: str
     description: str | None = None
     models: list[str] | None = None
@@ -38,9 +38,9 @@ class TagUpdateRequest(TagBase):
     budget_duration: str | None = None
 
 
-class TagDeleteRequest(BaseModel):
+class TagDeleteRequest(LiteLLMBaseModel):
     name: str
 
 
-class TagInfoRequest(BaseModel):
+class TagInfoRequest(LiteLLMBaseModel):
     names: list[str]
diff --git a/litellm/types/tool_management.py b/litellm/types/tool_management.py
index 13553dbecc6..5ea500fc0f2 100644
--- a/litellm/types/tool_management.py
+++ b/litellm/types/tool_management.py
@@ -5,7 +5,9 @@ Pydantic models for Tool Policy management endpoints.
 from datetime import datetime
 from typing import Literal
 
-from pydantic import BaseModel, Field
+from pydantic import Field
+
+from litellm.types.llms.base import LiteLLMBaseModel
 
 ToolCallPolicy = Literal["trusted", "untrusted", "dual_llm", "blocked"]
 
@@ -13,13 +15,13 @@ ToolInputPolicy = Literal["trusted", "untrusted", "blocked"]
 ToolOutputPolicy = Literal["trusted", "untrusted"]
 
 
-class ToolDiscoveryUser(BaseModel):
+class ToolDiscoveryUser(LiteLLMBaseModel):
     user_id: str
     user_email: str | None = None
     user_alias: str | None = None
 
 
-class LiteLLM_ToolTableRow(BaseModel):
+class LiteLLM_ToolTableRow(LiteLLMBaseModel):
     tool_id: str
     tool_name: str
     origin: str | None = None
@@ -39,12 +41,12 @@ class LiteLLM_ToolTableRow(BaseModel):
     updated_by: str | None = None
 
 
-class ToolListResponse(BaseModel):
+class ToolListResponse(LiteLLMBaseModel):
     tools: list[LiteLLM_ToolTableRow]
     total: int
 
 
-class ToolPolicyUpdateRequest(BaseModel):
+class ToolPolicyUpdateRequest(LiteLLMBaseModel):
     tool_name: str
     input_policy: ToolInputPolicy | None = None
     output_policy: ToolOutputPolicy | None = None
@@ -53,7 +55,7 @@ class ToolPolicyUpdateRequest(BaseModel):
     key_alias: str | None = None
 
 
-class ToolPolicyUpdateResponse(BaseModel):
+class ToolPolicyUpdateResponse(LiteLLMBaseModel):
     tool_name: str
     input_policy: ToolInputPolicy | None = None
     output_policy: ToolOutputPolicy | None = None
@@ -62,7 +64,7 @@ class ToolPolicyUpdateResponse(BaseModel):
     key_hash: str | None = None
 
 
-class ToolPolicyOverrideRow(BaseModel):
+class ToolPolicyOverrideRow(LiteLLMBaseModel):
     override_id: str
     tool_name: str
     team_id: str | None = None
@@ -73,23 +75,23 @@ class ToolPolicyOverrideRow(BaseModel):
     updated_at: datetime | None = None
 
 
-class ToolPolicyOption(BaseModel):
+class ToolPolicyOption(LiteLLMBaseModel):
     value: str
     label: str
     description: str
 
 
-class ToolPolicyOptionsResponse(BaseModel):
+class ToolPolicyOptionsResponse(LiteLLMBaseModel):
     input_policies: list[ToolPolicyOption]
     output_policies: list[ToolPolicyOption]
 
 
-class ToolDetailResponse(BaseModel):
+class ToolDetailResponse(LiteLLMBaseModel):
     tool: LiteLLM_ToolTableRow
     overrides: list[ToolPolicyOverrideRow] = Field(default_factory=list)
 
 
-class ToolUsageLogEntry(BaseModel):
+class ToolUsageLogEntry(LiteLLMBaseModel):
     """One spend log row for a tool call (for UI "recent logs" table)."""
 
     id: str  # request_id
@@ -100,14 +102,14 @@ class ToolUsageLogEntry(BaseModel):
     input_snippet: str | None = None
 
 
-class ToolUsageLogsResponse(BaseModel):
+class ToolUsageLogsResponse(LiteLLMBaseModel):
     logs: list[ToolUsageLogEntry]
     total: int
     page: int
     page_size: int
 
 
-class ToolSpendEntry(BaseModel):
+class ToolSpendEntry(LiteLLMBaseModel):
     """Total spend attributed to one tool over the requested window."""
 
     tool_name: str
@@ -119,7 +121,7 @@ class ToolSpendEntry(BaseModel):
     total_tokens: int = 0
 
 
-class ToolSpendDailyEntry(BaseModel):
+class ToolSpendDailyEntry(LiteLLMBaseModel):
     """Spend attributed to one tool on one UTC day."""
 
     date: str
@@ -128,7 +130,7 @@ class ToolSpendDailyEntry(BaseModel):
     call_count: int = 0
 
 
-class ToolSpendResponse(BaseModel):
+class ToolSpendResponse(LiteLLMBaseModel):
     by_tool: list[ToolSpendEntry] = Field(default_factory=list)
     daily: list[ToolSpendDailyEntry] = Field(default_factory=list)
     start_date: str | None = None
diff --git a/litellm/types/utils.py b/litellm/types/utils.py
index 2c4dff592e5..d3fe65037e2 100644
--- a/litellm/types/utils.py
+++ b/litellm/types/utils.py
@@ -52,6 +52,7 @@ from litellm.constants import INTERNAL_KWARG_PREFIX
 from litellm.types.llms.base import (
     BaseLiteLLMOpenAIResponseObject,
     CachedTokensDetails,
+    LiteLLMBaseModel,
     LiteLLMPydanticObjectBase,
 )
 from litellm.types.mcp import MCPServerCostInfo
@@ -1740,7 +1741,7 @@ class CompletionTokensDetailsWrapper(CompletionTokensDetails):  # wrapper for ol
     """Video tokens generated by the model."""
 
 
-class CacheCreationTokenDetails(BaseModel):
+class CacheCreationTokenDetails(LiteLLMBaseModel):
     ephemeral_5m_input_tokens: int | None = None
     ephemeral_1h_input_tokens: int | None = None
 
@@ -1843,7 +1844,7 @@ class PromptTokensDetailsWrapper(
             del self.cached_tokens_details
 
 
-class ServerToolUse(BaseModel):
+class ServerToolUse(LiteLLMBaseModel):
     web_search_requests: int | None = None
     tool_search_requests: int | None = None
     browser_open_requests: int | None = None
@@ -2723,17 +2724,17 @@ class ImageResponse(OpenAIImageResponse, BaseLiteLLMOpenAIResponseObject):
             return self.dict()
 
 
-class TranscriptionUsageDurationObject(BaseModel):
+class TranscriptionUsageDurationObject(LiteLLMBaseModel):
     type: Literal["duration"]
     seconds: float
 
 
-class TranscriptionUsageInputTokenDetailsObject(BaseModel):
+class TranscriptionUsageInputTokenDetailsObject(LiteLLMBaseModel):
     audio_tokens: int
     text_tokens: int
 
 
-class TranscriptionUsageTokensObject(BaseModel):
+class TranscriptionUsageTokensObject(LiteLLMBaseModel):
     type: Literal["tokens"]
     input_tokens: int
     output_tokens: int
@@ -3708,7 +3709,7 @@ class StandardCallbackDynamicParams(TypedDict, total=False):
     litellm_disabled_callbacks: list[str] | None
 
 
-class MirroredPricingParams(BaseModel):
+class MirroredPricingParams(LiteLLMBaseModel):
     """Pricing overrides that ``Deployment.__init__`` mirrors from ``litellm_params``
     onto ``model_info``, so both blobs hold the same rate.
 
@@ -3997,7 +3998,7 @@ class StandardKeyGenerationConfig(TypedDict, total=False):
     personal_key_generation: PersonalUIKeyGenerationConfig
 
 
-class BudgetConfig(BaseModel):
+class BudgetConfig(LiteLLMBaseModel):
     max_budget: float | None = None
     budget_duration: str | None = None
     tpm_limit: int | None = None
@@ -4516,7 +4517,7 @@ class PriorityReservationDict(TypedDict, total=False):
     value: float
 
 
-class PriorityReservationSettings(BaseModel):
+class PriorityReservationSettings(LiteLLMBaseModel):
     """
     Settings for priority-based rate limiting reservation.
 
diff --git a/litellm/types/vector_stores.py b/litellm/types/vector_stores.py
index 34c14e6b042..5e7d517ece2 100644
--- a/litellm/types/vector_stores.py
+++ b/litellm/types/vector_stores.py
@@ -4,9 +4,10 @@ from datetime import datetime
 from enum import Enum
 from typing import Any, Literal
 
-from pydantic import BaseModel
 from typing_extensions import ReadOnly, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
+
 
 class SupportedVectorStoreIntegrations(str, Enum):
     """Supported vector store integrations."""
@@ -57,7 +58,7 @@ class LiteLLM_ManagedVectorStoreListResponse(TypedDict, total=False):
     total_pages: int | None
 
 
-class VectorStoreUpdateRequest(BaseModel):
+class VectorStoreUpdateRequest(LiteLLMBaseModel):
     vector_store_id: str
     custom_llm_provider: str | None = None
     vector_store_name: str | None = None
@@ -65,11 +66,11 @@ class VectorStoreUpdateRequest(BaseModel):
     vector_store_metadata: dict | None = None
 
 
-class VectorStoreDeleteRequest(BaseModel):
+class VectorStoreDeleteRequest(LiteLLMBaseModel):
     vector_store_id: str
 
 
-class VectorStoreInfoRequest(BaseModel):
+class VectorStoreInfoRequest(LiteLLMBaseModel):
     vector_store_id: str
 
 
@@ -262,12 +263,12 @@ class VectorStoreCreateResponse(TypedDict, total=False):
     metadata: dict[str, str] | None  # Metadata associated with the vector store
 
 
-class IndexCreateLiteLLMParams(BaseModel):
+class IndexCreateLiteLLMParams(LiteLLMBaseModel):
     vector_store_index: str
     vector_store_name: str
 
 
-class IndexCreateRequest(BaseModel):
+class IndexCreateRequest(LiteLLMBaseModel):
     index_name: str
     litellm_params: IndexCreateLiteLLMParams
     index_info: dict[str, object] | None = None
@@ -278,7 +279,7 @@ class BaseVectorStoreAuthCredentials(TypedDict, total=False):
     query_params: dict
 
 
-class LiteLLM_ManagedVectorStoreIndex(BaseModel):
+class LiteLLM_ManagedVectorStoreIndex(LiteLLMBaseModel):
     """LiteLLM managed vector store index object - this is is the object stored in the database"""
 
     id: str
@@ -291,7 +292,7 @@ class LiteLLM_ManagedVectorStoreIndex(BaseModel):
     updated_by: str | None = None
 
 
-class IndexListResponse(BaseModel):
+class IndexListResponse(LiteLLMBaseModel):
     object: Literal["list"] = "list"
     data: tuple[LiteLLM_ManagedVectorStoreIndex, ...]
 
diff --git a/litellm/types/videos/main.py b/litellm/types/videos/main.py
index f4369fd95af..7c22fe18745 100644
--- a/litellm/types/videos/main.py
+++ b/litellm/types/videos/main.py
@@ -2,11 +2,13 @@ import builtins
 from typing import Any, Literal
 
 from openai.types.audio.transcription_create_params import FileTypes
-from pydantic import BaseModel
+from pydantic import Field, PrivateAttr
 from typing_extensions import ReadOnly, TypedDict
 
+from litellm.types.llms.base import LiteLLMBaseModel
 
-class VideoObject(BaseModel):
+
+class VideoObject(LiteLLMBaseModel):
     """Represents a generated video object."""
 
     id: str
@@ -22,7 +24,7 @@ class VideoObject(BaseModel):
     size: str | None = None
     model: str | None = None
     usage: dict[str, Any] | None = None
-    _hidden_params: dict[str, builtins.object] = {}
+    _hidden_params: dict[str, builtins.object] = PrivateAttr(default={})
 
     def __contains__(self, key) -> bool:
         # Define custom behavior for the 'in' operator
@@ -44,11 +46,11 @@ class VideoObject(BaseModel):
             return self.dict()
 
 
-class VideoResponse(BaseModel):
+class VideoResponse(LiteLLMBaseModel):
     """Response object for video generation requests."""
 
     data: list[VideoObject]
-    hidden_params: dict[str, object] = {}
+    hidden_params: dict[str, object] = Field(default={})
 
     def __contains__(self, key) -> bool:
         return hasattr(self, key)
@@ -104,14 +106,14 @@ class DecodedVideoId(TypedDict, total=False):
     video_id: str
 
 
-class CharacterObject(BaseModel):
+class CharacterObject(LiteLLMBaseModel):
     """Represents a character created from a video."""
 
     id: str
     object: Literal["character"] = "character"
     created_at: int
     name: str
-    _hidden_params: dict[str, builtins.object] = {}
+    _hidden_params: dict[str, builtins.object] = PrivateAttr(default={})
 
     def __contains__(self, key) -> bool:
         return hasattr(self, key)
diff --git a/scripts/generate_trace_types.py b/scripts/generate_trace_types.py
index a8f4054a646..4ba462ae731 100644
--- a/scripts/generate_trace_types.py
+++ b/scripts/generate_trace_types.py
@@ -99,6 +99,8 @@ def generate(
         else (
             "--output-model-type",
             "pydantic_v2.BaseModel",
+            "--base-class",
+            "litellm.types.llms.base.LiteLLMBaseModel",
             "--enable-faux-immutability",
             "--additional-imports",
             "collections.abc.Mapping,typing.TypeAlias,pydantic.JsonValue",
diff --git a/tests/unit/types/llms/test_types_llms_base.py b/tests/unit/types/llms/test_types_llms_base.py
new file mode 100644
index 00000000000..32f22e68397
--- /dev/null
+++ b/tests/unit/types/llms/test_types_llms_base.py
@@ -0,0 +1,73 @@
+import os
+import subprocess
+import sys
+from typing import Final
+
+import pytest
+
+_PROBE: Final = """
+from litellm.types.llms.base import LiteLLMBaseModel
+
+
+class Probe(LiteLLMBaseModel):
+    value: int
+
+
+print(Probe.__pydantic_complete__, Probe(value=1).value, Probe.__pydantic_complete__)
+"""
+
+
+def _run_probe(probe: str, defer_pydantic_build: str | None) -> str:
+    env: Final = {key: value for key, value in os.environ.items() if key != "DEFER_PYDANTIC_BUILD"}
+    overrides: Final = {} if defer_pydantic_build is None else {"DEFER_PYDANTIC_BUILD": defer_pydantic_build}
+    result: Final = subprocess.run(
+        (sys.executable, "-I", "-c", probe),
+        env={**env, **overrides},
+        capture_output=True,
+        text=True,
+        check=True,
+        timeout=120,
+    )
+    return result.stdout.strip().splitlines()[-1]
+
+
+def _probe_with(defer_pydantic_build: str | None) -> str:
+    return _run_probe(_PROBE, defer_pydantic_build)
+
+
+@pytest.mark.parametrize("defer_pydantic_build", [None, "true", "1", "on"])
+def test_litellm_models_defer_schema_build_until_first_use(defer_pydantic_build: str | None) -> None:
+    assert _probe_with(defer_pydantic_build) == "False 1 True"
+
+
+@pytest.mark.parametrize("defer_pydantic_build", ["false", "0", "off"])
+def test_litellm_models_build_schema_at_class_creation_when_defer_is_disabled(defer_pydantic_build: str) -> None:
+    assert _probe_with(defer_pydantic_build) == "True 1 True"
+
+
+_NESTED_PROBE: Final = """
+from pydantic import BaseModel
+
+from litellm.types.llms.base import LiteLLMBaseModel
+
+
+class Inner(LiteLLMBaseModel):
+    value: int
+
+
+class Parent(LiteLLMBaseModel):
+    inner: Inner
+
+
+class Holder(BaseModel):
+    item: object
+
+
+validated = Parent(inner={"value": 1}).inner
+constructed = Inner.model_construct(value=2)
+print(Holder(item=validated).model_dump_json(serialize_as_any=True), Holder(item=constructed).model_dump_json())
+"""
+
+
+def test_deferred_instances_created_without_their_own_init_still_serialize_as_any() -> None:
+    assert _run_probe(_NESTED_PROBE, None) == '{"item":{"value":1}} {"item":{"value":2}}'