Merge remote-tracking branch 'origin/main' into litellm_migrate_tests_p11

litellm/a2a_protocol/providers/bedrock_agentcore/transformation.py

@@ -85,7 +85,7 @@ def _filter_reserved_headers(
 
 
 def _request_scoped_runtime_session_id(
-    params: Mapping[str, Any],
+    params: Mapping[str, object],
     litellm_params: Mapping[str, Any],
 ) -> str | None:
     context_id: Final = get_session_id_from_a2a_params(params)

litellm/a2a_protocol/providers/watsonx_orchestrate/config.py

@@ -20,7 +20,7 @@ class WatsonxOrchestrateA2AConfig(BaseA2AProviderConfig):
         params: dict[str, Any],
         api_base: str | None = None,
         **kwargs: Any,
-    ) -> dict[str, Any]:
+    ) -> dict[str, object]:
         """Handle a non-streaming A2A request via WXO runs API."""
         litellm_params: Final = kwargs.get("litellm_params")
         if not litellm_params:
@@ -40,7 +40,7 @@ class WatsonxOrchestrateA2AConfig(BaseA2AProviderConfig):
         params: dict[str, Any],
         api_base: str | None = None,
         **kwargs: Any,
-    ) -> AsyncIterator[dict[str, Any]]:
+    ) -> AsyncIterator[dict[str, object]]:
         """Handle a streaming A2A request via WXO streaming runs API."""
         litellm_params: Final = kwargs.get("litellm_params")
         if not litellm_params:

litellm/a2a_protocol/utils.py

@@ -17,7 +17,7 @@ class A2ARequestUtils:
     """Utility class for A2A request/response processing."""
 
     @staticmethod
-    def extract_text_from_message(message: Any) -> str:
+    def extract_text_from_message(message: object) -> str:
         """
         Extract text content from A2A message parts.
 
@@ -142,7 +142,7 @@ class A2ARequestUtils:
         return prompt_tokens, completion_tokens, total_tokens
 
 
-def get_session_id_from_a2a_params(params: Mapping[str, Any]) -> str | None:
+def get_session_id_from_a2a_params(params: Mapping[str, object]) -> str | None:
     message: Final = params.get("message", {})
     if isinstance(message, dict):
         return message.get("contextId")
@@ -166,7 +166,7 @@ def scope_session_to_principal(session_id: str, principal: str | None) -> str:
 
 
 # Backwards compatibility aliases
-def extract_text_from_a2a_message(message: Any) -> str:
+def extract_text_from_a2a_message(message: object) -> str:
     return A2ARequestUtils.extract_text_from_message(message)
 
 

litellm/integrations/gitlab/gitlab_prompt_manager.py

@@ -200,8 +200,8 @@ class GitLabTemplateManager:
             metadata=metadata,
         )
 
-    def _parse_yaml_basic(self, yaml_str: str) -> dict[str, Any]:
-        result: Final[dict[str, Any]] = {}
+    def _parse_yaml_basic(self, yaml_str: str) -> dict[str, bool | int | float | str]:
+        result: Final[dict[str, bool | int | float | str]] = {}
         for line in yaml_str.split("\n"):
             line = line.strip()
             if ":" in line and not line.startswith("#"):

litellm/integrations/vantage/vantage_logger.py

@@ -59,7 +59,7 @@ class VantageLogger(FocusLogger):
                     raw_interval,
                 )
 
-        destination_config: Final[dict[str, Any]] = {}
+        destination_config: Final[dict[str, str]] = {}
         if resolved_api_key:
             destination_config["api_key"] = resolved_api_key
         if resolved_token:
@@ -93,7 +93,7 @@ class VantageLogger(FocusLogger):
 
         pod_lock_manager = None
         if proxy_logging_obj is not None:
-            writer: Final = getattr(proxy_logging_obj, "db_spend_update_writer", None)
+            writer: Final[object] = getattr(proxy_logging_obj, "db_spend_update_writer", None)
             if writer is not None:
                 pod_lock_manager = getattr(writer, "pod_lock_manager", None)
 

litellm/interactions/agents/http_handler.py

@@ -7,7 +7,7 @@ duplicated. BaseAgentsAPIConfig stays as pure transform code.
 """
 
 from collections.abc import Coroutine, Mapping
-from typing import Any, Final
+from typing import Final
 
 import httpx
 
@@ -38,7 +38,7 @@ class AgentsHTTPHandler(InteractionsHTTPHandler):
         name: str,
         litellm_params: GenericLiteLLMParams,
         logging_obj: LiteLLMLoggingObj,
-        extra_headers: dict[str, Any] | None = None,
+        extra_headers: dict[str, str] | None = None,
         extra_body: Mapping[str, object] | None = None,
         timeout: float | httpx.Timeout | None = None,
         client: HTTPHandler | None = None,
@@ -93,7 +93,7 @@ class AgentsHTTPHandler(InteractionsHTTPHandler):
         name: str,
         litellm_params: GenericLiteLLMParams,
         logging_obj: LiteLLMLoggingObj,
-        extra_headers: dict[str, Any] | None = None,
+        extra_headers: dict[str, str] | None = None,
         extra_body: Mapping[str, object] | None = None,
         timeout: float | httpx.Timeout | None = None,
         client: AsyncHTTPHandler | None = None,
@@ -141,7 +141,7 @@ class AgentsHTTPHandler(InteractionsHTTPHandler):
         agents_api_config: BaseAgentsAPIConfig,
         litellm_params: GenericLiteLLMParams,
         logging_obj: LiteLLMLoggingObj,
-        extra_headers: dict[str, Any] | None = None,
+        extra_headers: dict[str, str] | None = None,
         timeout: float | httpx.Timeout | None = None,
         client: HTTPHandler | None = None,
         _is_async: bool = False,
@@ -181,7 +181,7 @@ class AgentsHTTPHandler(InteractionsHTTPHandler):
         agents_api_config: BaseAgentsAPIConfig,
         litellm_params: GenericLiteLLMParams,
         logging_obj: LiteLLMLoggingObj,
-        extra_headers: dict[str, Any] | None = None,
+        extra_headers: dict[str, str] | None = None,
         timeout: float | httpx.Timeout | None = None,
         client: AsyncHTTPHandler | None = None,
     ) -> AgentListResponse:
@@ -216,7 +216,7 @@ class AgentsHTTPHandler(InteractionsHTTPHandler):
         name: str,
         litellm_params: GenericLiteLLMParams,
         logging_obj: LiteLLMLoggingObj,
-        extra_headers: dict[str, Any] | None = None,
+        extra_headers: dict[str, str] | None = None,
         timeout: float | httpx.Timeout | None = None,
         client: HTTPHandler | None = None,
         _is_async: bool = False,
@@ -259,7 +259,7 @@ class AgentsHTTPHandler(InteractionsHTTPHandler):
         name: str,
         litellm_params: GenericLiteLLMParams,
         logging_obj: LiteLLMLoggingObj,
-        extra_headers: dict[str, Any] | None = None,
+        extra_headers: dict[str, str] | None = None,
         timeout: float | httpx.Timeout | None = None,
         client: AsyncHTTPHandler | None = None,
     ) -> AgentCreateResponse:
@@ -295,7 +295,7 @@ class AgentsHTTPHandler(InteractionsHTTPHandler):
         name: str,
         litellm_params: GenericLiteLLMParams,
         logging_obj: LiteLLMLoggingObj,
-        extra_headers: dict[str, Any] | None = None,
+        extra_headers: dict[str, str] | None = None,
         timeout: float | httpx.Timeout | None = None,
         client: HTTPHandler | None = None,
         _is_async: bool = False,
@@ -338,7 +338,7 @@ class AgentsHTTPHandler(InteractionsHTTPHandler):
         name: str,
         litellm_params: GenericLiteLLMParams,
         logging_obj: LiteLLMLoggingObj,
-        extra_headers: dict[str, Any] | None = None,
+        extra_headers: dict[str, str] | None = None,
         timeout: float | httpx.Timeout | None = None,
         client: AsyncHTTPHandler | None = None,
     ) -> AgentDeleteResult:
@@ -374,7 +374,7 @@ class AgentsHTTPHandler(InteractionsHTTPHandler):
         name: str,
         litellm_params: GenericLiteLLMParams,
         logging_obj: LiteLLMLoggingObj,
-        extra_headers: dict[str, Any] | None = None,
+        extra_headers: dict[str, str] | None = None,
         timeout: float | httpx.Timeout | None = None,
         client: HTTPHandler | None = None,
         _is_async: bool = False,
@@ -417,7 +417,7 @@ class AgentsHTTPHandler(InteractionsHTTPHandler):
         name: str,
         litellm_params: GenericLiteLLMParams,
         logging_obj: LiteLLMLoggingObj,
-        extra_headers: dict[str, Any] | None = None,
+        extra_headers: dict[str, str] | None = None,
         timeout: float | httpx.Timeout | None = None,
         client: AsyncHTTPHandler | None = None,
     ) -> AgentVersionsResponse:

litellm/litellm_core_utils/logging_utils.py

@@ -67,7 +67,9 @@ def _truncate_base64_in_string(value: str) -> str:
     return _DATA_URI_RE.sub(_base64_data_uri_replacer, value)
 
 
-def _truncate_base64_in_value(value: Any) -> Any:
+def _truncate_base64_in_value(
+    value: str | dict[str, object] | list[object] | None,
+) -> str | dict[str, object] | list[object] | None:
     """Iteratively truncate base64 data URIs in a JSON-like value (str/list/dict).
 
     Uses an explicit stack instead of recursion to satisfy the project's

litellm/litellm_core_utils/url_utils.py

@@ -418,7 +418,7 @@ def _extract_redirect_url(response: httpx.Response, request_url: str) -> str:
     return str(httpx.URL(request_url).join(location))
 
 
-def safe_get(client: Any, url: str, **kwargs: Any) -> httpx.Response:
+def safe_get(client: _UrlFetcher, url: str, **kwargs: Any) -> httpx.Response:
     """
     Fetch a user-supplied URL with SSRF protection on every redirect hop.
 
@@ -461,7 +461,7 @@ def safe_get(client: Any, url: str, **kwargs: Any) -> httpx.Response:
     raise SSRFError("Too many redirects")
 
 
-async def async_safe_get(client: Any, url: str, **kwargs: Any) -> httpx.Response:
+async def async_safe_get(client: _AsyncUrlFetcher, url: str, **kwargs: Any) -> httpx.Response:
     """Async version of safe_get."""
     if not getattr(litellm, "user_url_validation", True):
         kwargs.setdefault("follow_redirects", True)

litellm/llms/anthropic/chat/handler.py

@@ -596,7 +596,7 @@ class ModelResponseIterator:
         self.reasoning_content_chunks: list[str] = []
 
         # Track server tool use inputs and results for code_interpreter_results
-        self._server_tool_inputs: dict[str, Any] = {}
+        self._server_tool_inputs: dict[str, object] = {}
         self.tool_results: list[dict[str, Any]] = []
         self._current_server_tool_id: str | None = None
         self._container_id: str | None = None

litellm/llms/azure/completion/handler.py

@@ -1,6 +1,7 @@
 from collections.abc import Callable
-from typing import Any, Final
+from typing import Final
 
+import httpx
 from openai import AsyncAzureOpenAI, AzureOpenAI
 
 from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
@@ -191,7 +192,7 @@ class AzureTextCompletion(BaseAzureLLM):
         model: str,
         api_base: str,
         data: dict,
-        timeout: Any,
+        timeout: float | httpx.Timeout | None,
         model_response: ModelResponse,
         logging_obj: LiteLLMLoggingObj,
         max_retries: int,
@@ -253,7 +254,7 @@ class AzureTextCompletion(BaseAzureLLM):
         api_version: str,
         data: dict,
         model: str,
-        timeout: Any,
+        timeout: float | httpx.Timeout | None,
         azure_ad_token: str | None = None,
         client=None,
         litellm_params: dict = {},
@@ -306,7 +307,7 @@ class AzureTextCompletion(BaseAzureLLM):
         api_version: str,
         data: dict,
         model: str,
-        timeout: Any,
+        timeout: float | httpx.Timeout | None,
         azure_ad_token: str | None = None,
         client=None,
         litellm_params: dict = {},

litellm/llms/azure_ai/ocr/document_intelligence/transformation.py

@@ -12,7 +12,7 @@ import asyncio
 import re
 import time
 from collections.abc import Mapping
-from typing import TYPE_CHECKING, Any, Final
+from typing import TYPE_CHECKING, Final
 from urllib.parse import quote
 
 import httpx
@@ -127,7 +127,7 @@ class AzureDocumentIntelligenceOCRConfig(BaseOCRConfig):
 
     def map_ocr_params(
         self,
-        non_default_params: dict,
+        non_default_params: Mapping[str, object],
         optional_params: dict,
         model: str,
     ) -> dict:
@@ -164,7 +164,7 @@ class AzureDocumentIntelligenceOCRConfig(BaseOCRConfig):
             raise UnsupportedParamsError(message=f"{e}", model=model, llm_provider="azure_ai") from e
 
     @staticmethod
-    def _normalize_pages_param(pages: Any) -> str:
+    def _normalize_pages_param(pages: object) -> str:
         """
         Convert a caller-provided `pages` value to Azure DI's query-string
         form. Azure expects 1-based page numbers, grammar: `^(\\d+(-\\d+)?)(,\\s*(\\d+(-\\d+)?))*$`.
@@ -412,7 +412,7 @@ class AzureDocumentIntelligenceOCRConfig(BaseOCRConfig):
             raise ValueError("Document URL is required")
 
         # Build Azure DI request
-        data: Final[dict[str, Any]] = {}
+        data: Final[dict[str, str]] = {}
 
         # Check if it's a data URI (base64)
         if document_url.startswith("data:"):

litellm/llms/bedrock/batches/transformation.py

@@ -2,7 +2,7 @@ import os
 import re
 import time
 from collections.abc import Mapping
-from typing import TYPE_CHECKING, Any, Final, Literal, cast
+from typing import TYPE_CHECKING, Final, Literal, cast
 
 from httpx import Headers, Response
 from pydantic import TypeAdapter, ValidationError
@@ -170,7 +170,7 @@ class BedrockBatchesConfig(BaseAWSLLM, BaseBatchesConfig):
         create_batch_data: CreateBatchRequest,
         optional_params: dict,
         litellm_params: dict,
-    ) -> dict[str, Any]:
+    ) -> dict[str, object]:
         """
         Transform the batch creation request to Bedrock format.
 
@@ -354,7 +354,7 @@ class BedrockBatchesConfig(BaseAWSLLM, BaseBatchesConfig):
         )
 
     @staticmethod
-    def _get_openai_compatible_batch_metadata(metadata: Any) -> dict[str, str]:
+    def _get_openai_compatible_batch_metadata(metadata: object) -> dict[str, str]:
         """
         OpenAI Batch metadata only accepts string values.
         """
@@ -379,7 +379,7 @@ class BedrockBatchesConfig(BaseAWSLLM, BaseBatchesConfig):
         batch_id: str,
         optional_params: dict,
         litellm_params: dict,
-    ) -> dict[str, Any]:
+    ) -> dict[str, object]:
         """
         Transform batch retrieval request for Bedrock.
 
@@ -523,7 +523,7 @@ class BedrockBatchesConfig(BaseAWSLLM, BaseBatchesConfig):
             )
 
         # Enrich metadata with useful Bedrock fields
-        enriched_metadata_raw: Final[dict[str, Any]] = {
+        enriched_metadata_raw: Final[dict[str, object]] = {
             "jobName": response_data.get("jobName"),
             "clientRequestToken": response_data.get("clientRequestToken"),
             "modelId": response_data.get("modelId"),

litellm/llms/bedrock/chat/invoke_transformations/amazon_twelvelabs_pegasus_transformation.py

@@ -110,7 +110,7 @@ class AmazonTwelveLabsPegasusConfig(AmazonInvokeConfig, BaseConfig):
         headers: dict,
     ) -> dict:
         input_prompt: Final = self._convert_messages_to_prompt(messages=messages)
-        request_data: Final[dict[str, Any]] = {"inputPrompt": input_prompt}
+        request_data: Final[dict[str, object]] = {"inputPrompt": input_prompt}
 
         media_source: Final = self._build_media_source(optional_params)
         if media_source is not None:

litellm/llms/bytez/chat/transformation.py

@@ -335,10 +335,10 @@ class BytezChatConfig(BaseConfig):
 
 
 class BytezCustomStreamWrapper(CustomStreamWrapper):
-    def chunk_creator(self, chunk: Any):
+    def chunk_creator(self, chunk: object):
         try:
             model_response: Final = self.model_response_creator()
-            response_obj: dict[str, Any] = {}
+            response_obj: dict[str, object] = {}
 
             response_obj = {
                 "text": chunk,
@@ -346,7 +346,7 @@ class BytezCustomStreamWrapper(CustomStreamWrapper):
                 "finish_reason": "",
             }
 
-            completion_obj: Final[dict[str, Any]] = {"content": chunk}
+            completion_obj: Final[dict[str, object]] = {"content": chunk}
 
             return self.return_processed_chunk_logic(
                 completion_obj=completion_obj,

litellm/llms/custom_httpx/aiohttp_handler.py

@@ -1,5 +1,5 @@
 import ssl
-from collections.abc import Callable
+from collections.abc import AsyncIterable, Callable, Iterable
 from typing import TYPE_CHECKING, Any, Final, cast
 
 import aiohttp
@@ -212,7 +212,7 @@ class BaseLLMAIOHTTPHandler:
         litellm_params: dict,
         stream: bool = False,
         files: dict | None = None,
-        content: Any = None,
+        content: str | bytes | Iterable[bytes] | AsyncIterable[bytes] | None = None,
         params: dict | None = None,
     ) -> httpx.Response:
         max_retry_on_unprocessable_entity_error: Final = provider_config.max_retry_on_unprocessable_entity_error

litellm/llms/deprecated_providers/aleph_alpha.py

@@ -146,7 +146,7 @@ class AlephAlphaConfig:
                 setattr(self.__class__, key, value)
 
     @classmethod
-    def get_config(cls):
+    def get_config(cls) -> dict[str, object]:
         return {
             k: v
             for k, v in cls.__dict__.items()

litellm/llms/lemonade/chat/transformation.py

@@ -170,7 +170,7 @@ class LemonadeChatConfig(OpenAILikeChatConfig):
         model: str,
         api_base: str | None = None,
         api_key: str | None = None,
-    ) -> Any:
+    ) -> dict[str, object]:
         if model.startswith("lemonade/"):
             model = model.split("/", 1)[1]
 

litellm/llms/openai/image_edit/transformation.py

@@ -66,7 +66,7 @@ class OpenAIImageEditConfig(BaseImageEditConfig):
     def _add_image_to_files(
         self,
         files_list: list[tuple[str, Any]],
-        image: Any,
+        image: object,
         field_name: str,
     ) -> None:
         """Add an image to the files list with appropriate content type"""

litellm/llms/runwayml/text_to_speech/transformation.py

@@ -78,7 +78,7 @@ class RunwayMLTextToSpeechConfig(BaseTextToSpeechConfig):
         aspeech: bool,
         api_base: str | None,
         api_key: str | None,
-        **kwargs: Any,
+        **kwargs: object,
     ) -> Union[
         "HttpxBinaryResponseContent",
         Coroutine[object, object, "HttpxBinaryResponseContent"],

litellm/llms/vertex_ai/files/transformation.py

@@ -651,7 +651,7 @@ def _openai_batch_jsonl_entry_to_vertex_embeddings_rows(
 
 def _openai_batch_jsonl_entry_to_vertex_rows(
     openai_entry: dict[str, Any],
-    map_openai_to_vertex_params: Callable[[dict[str, Any]], dict[str, Any]],
+    map_openai_to_vertex_params: Callable[[dict[str, Any]], dict[str, object]],
 ) -> tuple[Mapping[str, object], ...]:
     """
     Transforms a single OpenAI JSONL batch entry into the Vertex rows it maps to.
@@ -774,7 +774,7 @@ class _OpenAIToVertexBatchUploadStream(BaseFileUploadStream):
     def __init__(
         self,
         openai_file_content: FileTypes,
-        map_openai_to_vertex_params: Callable[[dict[str, Any]], dict[str, Any]],
+        map_openai_to_vertex_params: Callable[[dict[str, Any]], dict[str, object]],
     ) -> None:
         self._openai_file_content = openai_file_content
         self._map_openai_to_vertex_params = map_openai_to_vertex_params
@@ -948,7 +948,7 @@ class VertexAIFilesConfig(VertexBase, BaseFilesConfig):
     def _map_openai_to_vertex_params(
         self,
         openai_request_body: dict[str, Any],
-    ) -> dict[str, Any]:
+    ) -> dict[str, object]:
         """
         wrapper to call VertexGeminiConfig.map_openai_params
         """

litellm/llms/vertex_ai/gemini_embeddings/batch_embed_content_handler.py

@@ -3,7 +3,7 @@ Google AI Studio /batchEmbedContents Embeddings Endpoint
 """
 
 import json
-from typing import TYPE_CHECKING, Any, Final, Literal
+from typing import TYPE_CHECKING, Final, Literal
 
 import httpx
 
@@ -210,7 +210,7 @@ class GoogleBatchEmbeddings(VertexLLM):
             )
 
         ### TRANSFORMATION (sync path) ###
-        request_data: Any
+        request_data: VertexAIBatchEmbeddingsRequestBody | dict[str, object]
         if use_embed_content:
             resolved_files = {}
             if api_key:

litellm/llms/vertex_ai/vertex_ai_non_gemini.py

@@ -64,7 +64,7 @@ def _get_client_from_cache(client_cache_key: str):
     return litellm.in_memory_llm_clients_cache.get_cache(client_cache_key)
 
 
-def _set_client_in_cache(client_cache_key: str, vertex_llm_model: Any):
+def _set_client_in_cache(client_cache_key: str, vertex_llm_model: object):
     litellm.in_memory_llm_clients_cache.set_cache(
         key=client_cache_key,
         value=vertex_llm_model,

litellm/llms/watsonx/audio_transcription/transformation.py

@@ -4,7 +4,7 @@ Translates from OpenAI's `/v1/audio/transcriptions` to IBM WatsonX's `/ml/v1/aud
 WatsonX follows the OpenAI spec for audio transcription.
 """
 
-from typing import Any, Final
+from typing import Final
 
 from httpx import Response
 
@@ -124,7 +124,7 @@ class IBMWatsonXAudioTranscriptionConfig(IBMWatsonXMixin, OpenAIWhisperAudioTran
         }
 
         # Convert TypedDict to regular dict for AudioTranscriptionRequestData
-        form_data_dict: Final[dict[str, Any]] = dict(form_data)
+        form_data_dict: Final[dict[str, object]] = dict(form_data)
 
         return AudioTranscriptionRequestData(data=form_data_dict, files=files)
 

litellm/proxy/a2a/agent_card.py

@@ -10,7 +10,7 @@ and uses LiteLLM auth.
 import re
 from collections.abc import Mapping
 from copy import deepcopy
-from typing import Any, Final, Literal
+from typing import Final, Literal
 
 SupportedA2AVersion = Literal["0.3", "1.0"]
 
@@ -44,7 +44,7 @@ def normalize_protocol_version(version: object) -> SupportedA2AVersion | None:
     return next((supported for supported in SUPPORTED_A2A_PROTOCOL_VERSIONS if supported == major_minor), None)
 
 
-def resolve_served_protocol_version(card: Mapping[str, Any] | None) -> str:
+def resolve_served_protocol_version(card: Mapping[str, object] | None) -> str:
     """Return the validated protocol version an agent card pins, else the default."""
     normalized: Final = normalize_protocol_version(card.get("protocolVersion") if card else None)
     return normalized if normalized is not None else LITELLM_A2A_PROTOCOL_VERSION
@@ -53,7 +53,7 @@ def resolve_served_protocol_version(card: Mapping[str, Any] | None) -> str:
 # Security scheme exposed by the LiteLLM-fronted agent card. Always replaces
 # whatever upstream advertised — the client must authenticate to the proxy,
 # not the upstream agent.
-LITELLM_SECURITY_SCHEMES: Final[dict[str, dict[str, Any]]] = {
+LITELLM_SECURITY_SCHEMES: Final[dict[str, dict[str, str]]] = {
     "LiteLLMKey": {
         "type": "http",
         "scheme": "bearer",
@@ -112,7 +112,7 @@ _ALLOWED_TOP_LEVEL_KEYS: Final = {
     "url",
 }
 
-_DEFAULT_SKILLS: Final[list[dict[str, Any]]] = [
+_DEFAULT_SKILLS: Final[list[dict[str, str | list[str]]]] = [
     {
         "id": "chat",
         "name": "Chat",
@@ -129,7 +129,7 @@ _DEFAULT_MODES: Final[list[str]] = ["text"]
 _DEFAULT_AGENT_VERSION: Final = "1.0.0"
 
 
-def _filter_capabilities(upstream_capabilities: Any) -> dict[str, Any]:
+def _filter_capabilities(upstream_capabilities: object) -> dict[str, object]:
     """Return a capabilities dict containing only allowlisted, truthy keys."""
     if not isinstance(upstream_capabilities, dict):
         return {}
@@ -143,13 +143,13 @@ def _default_litellm_provider(proxy_base_url: str) -> dict[str, str]:
 
 
 def merge_agent_card(
-    upstream_card: Mapping[str, Any] | None,
+    upstream_card: Mapping[str, object] | None,
     *,
     proxy_url: str,
     proxy_base_url: str,
     name: str | None = None,
     description: str | None = None,
-) -> dict[str, Any]:
+) -> dict[str, object]:
     """
     Build the LiteLLM-fronted agent card.
 
@@ -169,7 +169,7 @@ def merge_agent_card(
         A dict suitable for serving as the proxy's agent card. Only keys in
         the v1.0 AgentCard schema (plus ``supportedInterfaces``) are emitted.
     """
-    base: Final[dict[str, Any]] = deepcopy(dict(upstream_card)) if upstream_card else {}
+    base: Final[dict[str, object]] = deepcopy(dict(upstream_card)) if upstream_card else {}
 
     # Keep the upstream ``url`` on the stored card: the runtime A2A
     # invocation path reads it from ``agent_card_params`` to know where to

litellm/proxy/client/credentials.py

@@ -1,3 +1,4 @@
+from collections.abc import Mapping
 from typing import Any, Final
 
 import requests
@@ -69,8 +70,8 @@ class CredentialsManagementClient:
     def create(
         self,
         credential_name: str,
-        credential_info: dict[str, Any],
-        credential_values: dict[str, Any],
+        credential_info: Mapping[str, object],
+        credential_values: Mapping[str, object],
         return_request: bool = False,
     ) -> dict[str, Any] | requests.Request:
         """

litellm/proxy/guardrails/guardrail_hooks/guardrails_ai/guardrails_ai.py

@@ -7,7 +7,7 @@
 
 import json
 import os
-from typing import TYPE_CHECKING, Any, Final, Literal, TypedDict
+from typing import TYPE_CHECKING, Final, Literal, TypedDict
 
 from fastapi import HTTPException
 
@@ -181,7 +181,7 @@ class GuardrailsAI(CustomGuardrail):
     ):  # raise exception if invalid, return a str for the user to receive - if rejected, or return a modified dictionary for passing into litellm
         return await self.process_input(data=data, call_type=call_type)
 
-    async def async_logging_hook(self, kwargs: dict, result: Any, call_type: str) -> tuple[dict, Any]:
+    async def async_logging_hook(self, kwargs: dict, result: object, call_type: str) -> tuple[dict, object]:
         if call_type == "acompletion" or call_type == "completion":
             kwargs = await self.process_input(data=kwargs, call_type=call_type)
 

litellm/proxy/guardrails/guardrail_hooks/singulr/singulr.py

@@ -24,7 +24,7 @@ from litellm.llms.custom_httpx.http_handler import (
 )
 from litellm.proxy._types import UserAPIKeyAuth
 from litellm.types.guardrails import GuardrailEventHooks
-from litellm.types.llms.openai import AllMessageValues
+from litellm.types.llms.openai import AllMessageValues, ChatCompletionToolCallChunk
 from litellm.types.proxy.guardrails.guardrail_hooks.base import (
     GuardrailConfigModel,
 )
@@ -36,7 +36,7 @@ from litellm.types.proxy.guardrails.guardrail_hooks.singulr import (
     ToolCall,
     ToolCallFunction,
 )
-from litellm.types.utils import CallTypes, GenericGuardrailAPIInputs
+from litellm.types.utils import CallTypes, ChatCompletionMessageToolCall, GenericGuardrailAPIInputs
 
 _DEFAULT_API_BASE: Final = "http://localhost:8003"
 _GUARD_ENDPOINT: Final = "/api/v1/ai-gateway/litellm-v2"
@@ -339,7 +339,7 @@ class SingulrGuardrail(CustomGuardrail):
         return inputs
 
     @staticmethod
-    def _build_tool_call(tool_call: Mapping[str, Any]) -> "ToolCall | None":
+    def _build_tool_call(tool_call: ChatCompletionToolCallChunk | ChatCompletionMessageToolCall) -> "ToolCall | None":
         tool_call_id: Final = tool_call.get("id")
         fun: Final = tool_call.get("function")
         if not tool_call_id or not fun:

litellm/proxy/policy_engine/pipeline_executor.py

@@ -8,7 +8,7 @@ pass/fail actions (allow, block, next, modify_response) and data forwarding.
 import copy
 import time
 from collections.abc import Callable, Mapping, Sequence
-from typing import TYPE_CHECKING, Any, Final, Literal, TypeVar
+from typing import TYPE_CHECKING, Final, Literal, TypeVar
 
 from pydantic import BaseModel
 
@@ -314,11 +314,11 @@ class PipelineExecutor:
         steps: list[PipelineStep],
         mode: str,
         data: dict,
-        user_api_key_dict: Any,
+        user_api_key_dict: "UserAPIKeyAuth",
         call_type: str,
         policy_name: str,
         raw_request_snapshot: dict | None = None,  # mutable-ok: same request-payload shape as data
-        streaming_chunks: list[Any] | None = None,  # mutable-ok: shared buffered-stream chunks, read per step
+        streaming_chunks: list[object] | None = None,  # mutable-ok: shared buffered-stream chunks, read per step
         endpoint_translation: "BaseTranslation | None" = None,
     ) -> PipelineExecutionResult:
         """
@@ -490,10 +490,10 @@ class PipelineExecutor:
         step: PipelineStep,
         mode: str,
         data: dict,
-        user_api_key_dict: Any,
+        user_api_key_dict: "UserAPIKeyAuth",
         call_type: str,
         raw_request_snapshot: dict | None = None,  # mutable-ok: same request-payload shape as data
-        streaming_chunks: list[Any] | None = None,  # mutable-ok: shared buffered-stream chunks, read per step
+        streaming_chunks: list[object] | None = None,  # mutable-ok: shared buffered-stream chunks, read per step
         endpoint_translation: "BaseTranslation | None" = None,
     ) -> tuple[
         Literal["pass", "fail", "error"],
@@ -722,7 +722,7 @@ def _extract_error_message(e: Exception) -> str:
     if isinstance(e, ModifyResponseException):
         return str(e)
     if HTTPException is not None and isinstance(e, HTTPException):
-        detail: Final = getattr(e, "detail", None)
+        detail: Final[object] = getattr(e, "detail", None)
         if detail:
             return str(detail)
     return str(e)

litellm/router_strategy/adaptive_router/hooks.py

@@ -86,7 +86,7 @@ def _resolve_session_key(kwargs: dict[str, Any]) -> str | None:
     return hashlib.sha256(payload.encode("utf-8")).hexdigest()
 
 
-def _last_user_content(messages: list[dict[str, Any]] | None) -> str | None:
+def _last_user_content(messages: Sequence[Mapping[str, object]] | None) -> str | None:
     if not messages:
         return None
     for msg in reversed(messages):

litellm/router_strategy/auto_router/auto_router.py

@@ -3,7 +3,7 @@ Auto-Routing Strategy that works with a Semantic Router Config
 """
 
 import asyncio
-from collections.abc import Mapping
+from collections.abc import Mapping, Sequence
 from typing import TYPE_CHECKING, Any, Final, Optional
 
 from pydantic import BaseModel, ConfigDict
@@ -158,7 +158,7 @@ class AutoRouter(CustomLogger):
         return await asyncio.shield(build_task)
 
     @staticmethod
-    def _extract_text_from_messages(messages: list[dict[str, Any]]) -> str:
+    def _extract_text_from_messages(messages: Sequence[Mapping[str, object]]) -> str:
         """
         Extract text content from the last user message for routing.
 

tests/test_litellm/llms/bedrock/image/test_bedrock_image_bearer_token.py

@@ -1,158 +0,0 @@
-import json
-import os
-from unittest.mock import Mock, patch
-import pytest
-
-
-import litellm
-from litellm.llms.custom_httpx.http_handler import HTTPHandler, AsyncHTTPHandler
-
-# Mock response for Bedrock image generation
-mock_image_response = {"images": ["base64_encoded_image_data"], "error": None}
-
-
-class TestBedrockImageGeneration:
-    def test_image_generation_with_api_key_bearer_token(self):
-        """Test image generation with bearer token authentication"""
-        test_api_key = "test-bearer-token-12345"
-        model = "bedrock/stability.sd3-large-v1:0"
-        prompt = "A cute baby sea otter"
-
-        with patch(
-            "litellm.llms.bedrock.image_generation.image_handler.BedrockImageGeneration.image_generation"
-        ) as mock_bedrock_image_gen:
-            # Setup mock response
-            mock_image_response_obj = litellm.ImageResponse()
-            mock_image_response_obj.data = [{"url": "https://example.com/image.jpg"}]
-            mock_bedrock_image_gen.return_value = mock_image_response_obj
-
-            response = litellm.image_generation(
-                model=model,
-                prompt=prompt,
-                aws_region_name="us-west-2",
-                api_key=test_api_key,
-            )
-
-            assert response is not None
-            assert len(response.data) > 0
-
-            mock_bedrock_image_gen.assert_called_once()
-            for call in mock_bedrock_image_gen.call_args_list:
-                if "headers" in call.kwargs:
-                    headers = call.kwargs["headers"]
-                    if (
-                        "Authorization" in headers
-                        and headers["Authorization"] == f"Bearer {test_api_key}"
-                    ):
-                        break
-
-    def test_image_generation_with_env_variable_bearer_token(self, monkeypatch):
-        """Test image generation with bearer token from environment variable"""
-        test_api_key = "env-bearer-token-12345"
-        model = "bedrock/stability.sd3-large-v1:0"
-        prompt = "A cute baby sea otter"
-
-        # Mock the environment variable
-        with (
-            patch.dict(os.environ, {"AWS_BEARER_TOKEN_BEDROCK": test_api_key}),
-            patch(
-                "litellm.llms.bedrock.image_generation.image_handler.BedrockImageGeneration.image_generation"
-            ) as mock_bedrock_image_gen,
-        ):
-
-            mock_image_response_obj = litellm.ImageResponse()
-            mock_image_response_obj.data = [{"url": "https://example.com/image.jpg"}]
-            mock_bedrock_image_gen.return_value = mock_image_response_obj
-
-            response = litellm.image_generation(
-                model=model, prompt=prompt, aws_region_name="us-west-2"
-            )
-
-            assert response is not None
-            assert len(response.data) > 0
-
-            mock_bedrock_image_gen.assert_called_once()
-            for call in mock_bedrock_image_gen.call_args_list:
-                if "headers" in call.kwargs:
-                    headers = call.kwargs["headers"]
-                    if (
-                        "Authorization" in headers
-                        and headers["Authorization"] == f"Bearer {test_api_key}"
-                    ):
-                        break
-
-    @pytest.mark.asyncio
-    async def test_async_image_generation_with_bearer_token(self):
-        """Test async image generation with bearer token authentication"""
-        test_api_key = "async-bearer-token-12345"
-        model = "bedrock/stability.sd3-large-v1:0"
-        prompt = "A cute baby sea otter"
-
-        with patch(
-            "litellm.llms.bedrock.image_generation.image_handler.BedrockImageGeneration.async_image_generation"
-        ) as mock_async_bedrock_image_gen:
-            mock_image_response_obj = litellm.ImageResponse()
-            mock_image_response_obj.data = [{"url": "https://example.com/image.jpg"}]
-            mock_async_bedrock_image_gen.return_value = mock_image_response_obj
-
-            # Call async image generation with api_key parameter
-            response = await litellm.aimage_generation(
-                model=model,
-                prompt=prompt,
-                aws_region_name="us-west-2",
-                api_key=test_api_key,
-            )
-
-            assert response is not None
-            assert len(response.data) > 0
-
-            mock_async_bedrock_image_gen.assert_called_once()
-            for call in mock_async_bedrock_image_gen.call_args_list:
-                if "headers" in call.kwargs:
-                    headers = call.kwargs["headers"]
-                    if (
-                        "Authorization" in headers
-                        and headers["Authorization"] == f"Bearer {test_api_key}"
-                    ):
-                        break
-
-    def test_image_generation_with_sigv4(self):
-        """Test image generation falls back to SigV4 auth when no bearer token is provided"""
-        model = "bedrock/stability.sd3-large-v1:0"
-        prompt = "A cute baby sea otter"
-
-        with patch(
-            "litellm.llms.bedrock.image_generation.image_handler.BedrockImageGeneration.image_generation"
-        ) as mock_bedrock_image_gen:
-            mock_image_response_obj = litellm.ImageResponse()
-            mock_image_response_obj.data = [{"url": "https://example.com/image.jpg"}]
-            mock_bedrock_image_gen.return_value = mock_image_response_obj
-
-            response = litellm.image_generation(
-                model=model, prompt=prompt, aws_region_name="us-west-2"
-            )
-
-            assert response is not None
-            assert len(response.data) > 0
-            mock_bedrock_image_gen.assert_called_once()
-
-
-def test_image_generation_bearer_token_never_runs_the_sigv4_credential_chain(monkeypatch):
-    """The deployment's AWS profile does not exist, so resolving SigV4 credentials
-    raises; a bearer-token deployment must still sign the request with the
-    bearer token alone."""
-    from litellm.llms.bedrock.image_generation.image_handler import BedrockImageGeneration
-
-    monkeypatch.setenv("AWS_BEARER_TOKEN_BEDROCK", "env-bearer-token-12345")
-
-    request = BedrockImageGeneration()._prepare_request(
-        model="amazon.nova-canvas-v1:0",
-        prompt="A cute baby sea otter",
-        optional_params={"aws_region_name": "us-west-2", "aws_profile_name": "litellm-no-such-aws-profile"},
-        api_base=None,
-        extra_headers=None,
-        api_key=None,
-        logging_obj=Mock(),
-    )
-
-    assert request.prepped.headers["Authorization"] == "Bearer env-bearer-token-12345"

tests/test_litellm/llms/hosted_vllm/chat/test_hosted_vllm_ssl_verify.py

@@ -1,147 +0,0 @@
-"""
-Test SSL verification for hosted_vllm provider.
-
-This test ensures that the ssl_verify parameter is properly passed through
-to the HTTP client when using the hosted_vllm provider.
-
-Issue: ssl_verify parameter was being ignored because hosted_vllm fell through
-to the OpenAI catch-all path in main.py, which doesn't pass ssl_verify to the HTTP client.
-"""
-
-from unittest.mock import MagicMock, patch
-
-import pytest
-
-
-import litellm
-
-
-class TestHostedVLLMSSLVerify:
-    """Test suite for SSL verification in hosted_vllm provider."""
-
-    @patch("litellm.llms.custom_httpx.llm_http_handler._get_httpx_client")
-    def test_hosted_vllm_ssl_verify_false_sync(self, mock_get_httpx_client):
-        """Test that ssl_verify=False is passed to the HTTP client for sync calls."""
-        # Setup mock client
-        mock_client = MagicMock()
-        mock_response = MagicMock()
-        mock_response.status_code = 200
-        mock_response.headers = {"content-type": "application/json"}
-        mock_response.json.return_value = {
-            "id": "chatcmpl-test",
-            "object": "chat.completion",
-            "created": 1234567890,
-            "model": "test-model",
-            "choices": [
-                {
-                    "index": 0,
-                    "message": {
-                        "role": "assistant",
-                        "content": "Test response",
-                    },
-                    "finish_reason": "stop",
-                }
-            ],
-            "usage": {
-                "prompt_tokens": 10,
-                "completion_tokens": 5,
-                "total_tokens": 15,
-            },
-        }
-        mock_response.text = '{"id": "chatcmpl-test", "object": "chat.completion", "created": 1234567890, "model": "test-model", "choices": [{"index": 0, "message": {"role": "assistant", "content": "Test response"}, "finish_reason": "stop"}], "usage": {"prompt_tokens": 10, "completion_tokens": 5, "total_tokens": 15}}'
-        mock_client.post.return_value = mock_response
-        mock_get_httpx_client.return_value = mock_client
-
-        try:
-            litellm.completion(
-                model="hosted_vllm/test-model",
-                messages=[{"role": "user", "content": "Hello"}],
-                api_base="https://test-vllm.example.com/v1",
-                ssl_verify=False,
-            )
-        except Exception:
-            # Even if the response parsing fails, we just need to verify
-            # that the mock was called with the correct ssl_verify parameter
-            pass
-
-        # Verify _get_httpx_client was called with ssl_verify=False
-        mock_get_httpx_client.assert_called()
-        call_args = mock_get_httpx_client.call_args
-
-        # Check that params contains ssl_verify=False
-        if call_args[0]:
-            # Positional argument
-            params = call_args[0][0]
-        else:
-            # Keyword argument
-            params = call_args[1].get("params", {})
-
-        assert (
-            params.get("ssl_verify") is False
-        ), f"Expected ssl_verify=False in params, got {params}"
-
-    @patch("litellm.llms.custom_httpx.llm_http_handler.get_async_httpx_client")
-    @pytest.mark.asyncio
-    async def test_hosted_vllm_ssl_verify_false_async(
-        self, mock_get_async_httpx_client
-    ):
-        """Test that ssl_verify=False is passed to the HTTP client for async calls."""
-        # Setup mock async client
-        mock_client = MagicMock()
-        mock_response = MagicMock()
-        mock_response.status_code = 200
-        mock_response.headers = {"content-type": "application/json"}
-        mock_response.json.return_value = {
-            "id": "chatcmpl-test",
-            "object": "chat.completion",
-            "created": 1234567890,
-            "model": "test-model",
-            "choices": [
-                {
-                    "index": 0,
-                    "message": {
-                        "role": "assistant",
-                        "content": "Test response",
-                    },
-                    "finish_reason": "stop",
-                }
-            ],
-            "usage": {
-                "prompt_tokens": 10,
-                "completion_tokens": 5,
-                "total_tokens": 15,
-            },
-        }
-        mock_response.text = '{"id": "chatcmpl-test", "object": "chat.completion", "created": 1234567890, "model": "test-model", "choices": [{"index": 0, "message": {"role": "assistant", "content": "Test response"}, "finish_reason": "stop"}], "usage": {"prompt_tokens": 10, "completion_tokens": 5, "total_tokens": 15}}'
-
-        async def mock_post(*args, **kwargs):
-            return mock_response
-
-        mock_client.post = mock_post
-        mock_get_async_httpx_client.return_value = mock_client
-
-        try:
-            await litellm.acompletion(
-                model="hosted_vllm/test-model",
-                messages=[{"role": "user", "content": "Hello"}],
-                api_base="https://test-vllm.example.com/v1",
-                ssl_verify=False,
-            )
-        except Exception:
-            # Even if the response parsing fails, we just need to verify
-            # that the mock was called with the correct ssl_verify parameter
-            pass
-
-        # Verify get_async_httpx_client was called with ssl_verify=False
-        mock_get_async_httpx_client.assert_called()
-        call_kwargs = mock_get_async_httpx_client.call_args[1]
-
-        # Check that params contains ssl_verify=False
-        params = call_kwargs.get("params", {})
-        assert (
-            params.get("ssl_verify") is False
-        ), f"Expected ssl_verify=False in params, got {params}"
-
-
-if __name__ == "__main__":
-    pytest.main([__file__, "-v", "-s"])

tests/test_litellm/llms/hosted_vllm/embedding/test_hosted_vllm_embedding_ssl_verify.py

@@ -1,135 +0,0 @@
-"""
-Test SSL verification for hosted_vllm provider embeddings.
-
-This test ensures that the ssl_verify parameter is properly passed through
-to the HTTP client when using the hosted_vllm provider for embeddings.
-
-Issue: ssl_verify parameter was being ignored because hosted_vllm fell through
-to the openai_like catch-all path in main.py, which doesn't pass ssl_verify to the HTTP client.
-"""
-
-from unittest.mock import MagicMock, patch
-
-import pytest
-
-
-import litellm
-
-
-class TestHostedVLLMEmbeddingSSLVerify:
-    """Test suite for SSL verification in hosted_vllm provider embeddings."""
-
-    @patch("litellm.llms.custom_httpx.llm_http_handler._get_httpx_client")
-    def test_hosted_vllm_embedding_ssl_verify_false_sync(self, mock_get_httpx_client):
-        """Test that ssl_verify=False is passed to the HTTP client for sync embedding calls."""
-        # Setup mock client
-        mock_client = MagicMock()
-        mock_response = MagicMock()
-        mock_response.status_code = 200
-        mock_response.headers = {"content-type": "application/json"}
-        mock_response.json.return_value = {
-            "object": "list",
-            "data": [
-                {
-                    "object": "embedding",
-                    "index": 0,
-                    "embedding": [0.1, 0.2, 0.3, 0.4, 0.5],
-                }
-            ],
-            "model": "text-embedding-model",
-            "usage": {
-                "prompt_tokens": 5,
-                "total_tokens": 5,
-            },
-        }
-        mock_response.text = '{"object": "list", "data": [{"object": "embedding", "index": 0, "embedding": [0.1, 0.2, 0.3, 0.4, 0.5]}], "model": "text-embedding-model", "usage": {"prompt_tokens": 5, "total_tokens": 5}}'
-        mock_client.post.return_value = mock_response
-        mock_get_httpx_client.return_value = mock_client
-
-        try:
-            litellm.embedding(
-                model="hosted_vllm/text-embedding-model",
-                input=["hello world"],
-                api_base="https://test-vllm.example.com/v1",
-                ssl_verify=False,
-            )
-        except Exception:
-            # Even if the response parsing fails, we just need to verify
-            # that the mock was called with the correct ssl_verify parameter
-            pass
-
-        # Verify _get_httpx_client was called with ssl_verify=False
-        mock_get_httpx_client.assert_called()
-        call_args = mock_get_httpx_client.call_args
-
-        # Check that params contains ssl_verify=False
-        if call_args[0]:
-            # Positional argument
-            params = call_args[0][0]
-        else:
-            # Keyword argument
-            params = call_args[1].get("params", {})
-
-        assert (
-            params.get("ssl_verify") is False
-        ), f"Expected ssl_verify=False in params, got {params}"
-
-    @patch("litellm.llms.custom_httpx.llm_http_handler.get_async_httpx_client")
-    @pytest.mark.asyncio
-    async def test_hosted_vllm_embedding_ssl_verify_false_async(
-        self, mock_get_async_httpx_client
-    ):
-        """Test that ssl_verify=False is passed to the HTTP client for async embedding calls."""
-        # Setup mock async client
-        mock_client = MagicMock()
-        mock_response = MagicMock()
-        mock_response.status_code = 200
-        mock_response.headers = {"content-type": "application/json"}
-        mock_response.json.return_value = {
-            "object": "list",
-            "data": [
-                {
-                    "object": "embedding",
-                    "index": 0,
-                    "embedding": [0.1, 0.2, 0.3, 0.4, 0.5],
-                }
-            ],
-            "model": "text-embedding-model",
-            "usage": {
-                "prompt_tokens": 5,
-                "total_tokens": 5,
-            },
-        }
-        mock_response.text = '{"object": "list", "data": [{"object": "embedding", "index": 0, "embedding": [0.1, 0.2, 0.3, 0.4, 0.5]}], "model": "text-embedding-model", "usage": {"prompt_tokens": 5, "total_tokens": 5}}'
-
-        async def mock_post(*args, **kwargs):
-            return mock_response
-
-        mock_client.post = mock_post
-        mock_get_async_httpx_client.return_value = mock_client
-
-        try:
-            await litellm.aembedding(
-                model="hosted_vllm/text-embedding-model",
-                input=["hello world"],
-                api_base="https://test-vllm.example.com/v1",
-                ssl_verify=False,
-            )
-        except Exception:
-            # Even if the response parsing fails, we just need to verify
-            # that the mock was called with the correct ssl_verify parameter
-            pass
-
-        # Verify get_async_httpx_client was called with ssl_verify=False
-        mock_get_async_httpx_client.assert_called()
-        call_kwargs = mock_get_async_httpx_client.call_args[1]
-
-        # Check that params contains ssl_verify=False
-        params = call_kwargs.get("params", {})
-        assert (
-            params.get("ssl_verify") is False
-        ), f"Expected ssl_verify=False in params, got {params}"
-
-
-if __name__ == "__main__":
-    pytest.main([__file__, "-v", "-s"])

tests/unit/llms/bedrock/image/test_bedrock_image_bearer_token.py

@@ -0,0 +1,21 @@
+from unittest.mock import Mock
+
+def test_image_generation_bearer_token_never_runs_the_sigv4_credential_chain(monkeypatch):
+    """The deployment's AWS profile does not exist, so resolving SigV4 credentials
+    raises; a bearer-token deployment must still sign the request with the
+    bearer token alone."""
+    from litellm.llms.bedrock.image_generation.image_handler import BedrockImageGeneration
+
+    monkeypatch.setenv("AWS_BEARER_TOKEN_BEDROCK", "env-bearer-token-12345")
+
+    request = BedrockImageGeneration()._prepare_request(
+        model="amazon.nova-canvas-v1:0",
+        prompt="A cute baby sea otter",
+        optional_params={"aws_region_name": "us-west-2", "aws_profile_name": "litellm-no-such-aws-profile"},
+        api_base=None,
+        extra_headers=None,
+        api_key=None,
+        logging_obj=Mock(),
+    )
+
+    assert request.prepped.headers["Authorization"] == "Bearer env-bearer-token-12345"

tests/unit/llms/bedrock/image/test_bedrock_image_prepare_request.py

@@ -11,7 +11,8 @@ def test_bedrock_image_prepare_request_with_arn() -> None:
 
     with (
         patch(
-            "litellm.llms.bedrock.image_generation.image_handler.BedrockImageGeneration._get_boto_credentials_from_optional_params"
+            "litellm.llms.bedrock.image_generation.image_handler.BedrockImageGeneration."
+            "_get_boto_credentials_from_optional_params"
         ),
         patch(
             "litellm.llms.bedrock.image_generation.image_handler.BedrockImageGeneration.get_request_headers"
@@ -31,7 +32,8 @@ def test_bedrock_image_prepare_request_with_arn() -> None:
 
     assert (
         request.endpoint_url
-        == "https://bedrock-runtime.test.com/model/arn%3Aaws%3Abedrock%3Aus-east-1%3A123456789012%3Aapplication-inference-profile%2Fabcdefghi123/invoke"
+        == "https://bedrock-runtime.test.com/model/arn%3Aaws%3Abedrock%3Aus-east-1%3A123456789012"
+        "%3Aapplication-inference-profile%2Fabcdefghi123/invoke"
     )
 
 
@@ -41,7 +43,8 @@ def test_bedrock_image_prepare_request_without_arn() -> None:
 
     with (
         patch(
-            "litellm.llms.bedrock.image_generation.image_handler.BedrockImageGeneration._get_boto_credentials_from_optional_params"
+            "litellm.llms.bedrock.image_generation.image_handler.BedrockImageGeneration."
+            "_get_boto_credentials_from_optional_params"
         ),
         patch(
             "litellm.llms.bedrock.image_generation.image_handler.BedrockImageGeneration.get_request_headers"

tests/unit/llms/bedrock/passthrough/guardrail_translation/test_handler.py

@@ -1072,7 +1072,8 @@ class TestDeAnonymizeConverseStream:
 
     @pytest.mark.asyncio
     async def test_reasoning_text_delta_de_anonymized(self):
-        """Reasoning deltas carry model output; their text must be guardrailed while the reasoning signature is left untouched."""
+        """Reasoning deltas carry model output; their text must be guardrailed while the
+        reasoning signature is left untouched."""
         stream_bytes = (
             _build_event_stream_frame("messageStart", {"role": "assistant"})
             + _build_event_stream_frame(
@@ -1105,7 +1106,8 @@ class TestDeAnonymizeConverseStream:
 
     @pytest.mark.asyncio
     async def test_tool_use_input_delta_de_anonymized(self):
-        """toolUse.input deltas carry model-generated tool arguments and must be guardrailed instead of being forwarded raw."""
+        """toolUse.input deltas carry model-generated tool arguments and must be
+        guardrailed instead of being forwarded raw."""
         stream_bytes = _build_event_stream_frame(
             "contentBlockDelta",
             {"contentBlockIndex": 0, "delta": {"toolUse": {"input": '{"q":"<PERSON_1>"}'}}},
@@ -1154,7 +1156,8 @@ class TestDeAnonymizeConverseStream:
 
     @pytest.mark.asyncio
     async def test_text_and_reasoning_deltas_de_anonymized_independently(self):
-        """Distinct delta kinds must each be guardrailed and written back into their own field without bleeding the de-anonymized text across kinds."""
+        """Distinct delta kinds must each be guardrailed and written back into their own
+        field without bleeding the de-anonymized text across kinds."""
         captured = {}
 
         async def mock_hook(data, user_api_key_dict, response):
@@ -1192,7 +1195,8 @@ class TestDeAnonymizeConverseStream:
 
     @pytest.mark.asyncio
     async def test_reasoning_signature_only_frame_left_unmodified(self):
-        """A reasoning delta carrying only a signature has no guardrailable text; it must be forwarded untouched and the guardrail must not run."""
+        """A reasoning delta carrying only a signature has no guardrailable text; it must
+        be forwarded untouched and the guardrail must not run."""
         stream_bytes = _build_event_stream_frame(
             "contentBlockDelta",
             {"contentBlockIndex": 0, "delta": {"reasoningContent": {"signature": "sig"}}},

tests/unit/llms/bedrock/passthrough/test_bedrock_passthrough_transformation.py

@@ -367,8 +367,6 @@ def test_bedrock_passthrough_region_extraction_from_inference_profile_arn():
         assert (
             "us-west-2" in api_base
         ), f"Expected region 'us-west-2' from ARN in base URL, but got: {api_base}"
-
-
 def test_bedrock_passthrough_model_id_arn_encoding():
     """
     Test that model_id ARNs are properly URL-encoded when used in endpoints.
@@ -421,7 +419,9 @@ def test_bedrock_passthrough_model_id_arn_encoding():
         ), f"ARN slash should be encoded, but found unencoded version in: {url_str}"
 
         # Verify the complete expected URL structure
-        expected_encoded_model_id = "arn:aws:bedrock:us-east-1:590183661440:application-inference-profile%2Fb943q2qbl3m7"
+        expected_encoded_model_id = (
+            "arn:aws:bedrock:us-east-1:590183661440:application-inference-profile%2Fb943q2qbl3m7"
+        )
         expected_url = f"https://bedrock-runtime.us-east-1.amazonaws.com/model/{expected_encoded_model_id}/converse"
         assert url_str == expected_url, f"Expected {expected_url}, but got: {url_str}"
 
@@ -517,7 +517,10 @@ def test_bedrock_passthrough_model_id_without_arn():
 def _event_frame(event_type: str, payload: dict) -> bytes:
     def header(name: str, value: str) -> bytes:
         name_b, value_b = name.encode(), value.encode()
-        return struct.pack("!B", len(name_b)) + name_b + struct.pack("!B", 7) + struct.pack("!H", len(value_b)) + value_b
+        return (
+            struct.pack("!B", len(name_b)) + name_b
+            + struct.pack("!B", 7) + struct.pack("!H", len(value_b)) + value_b
+        )
 
     payload_b = json.dumps(payload, separators=(",", ":")).encode()
     headers_b = (
@@ -591,7 +594,9 @@ def _feed(collector: PassthroughStreamCollector, stream: bytes, chunk_size: int
 
 def test_converse_stream_collector_keeps_usage_without_retaining_the_stream():
     texts = [f"tok{i} " for i in range(4000)]
-    stream = _event_frame("messageStart", {"role": "assistant"}) + _text_block(0, texts) + _stream_tail("end_turn", 4000)
+    stream = (
+        _event_frame("messageStart", {"role": "assistant"}) + _text_block(0, texts) + _stream_tail("end_turn", 4000)
+    )
     _feed(_converse_stream_collector(), stream)
 
     tracemalloc.start()

tests/unit/llms/bedrock/realtime/test_bedrock_realtime_handler.py

@@ -18,6 +18,21 @@ from litellm.llms.bedrock.realtime.handler import BedrockRealtime
 from litellm.llms.bedrock.realtime.transformation import BedrockRealtimeConfig
 
 
+@pytest.fixture(autouse=True)
+def _isolate_host_aws_config(monkeypatch, tmp_path):
+    monkeypatch.setenv("AWS_SHARED_CREDENTIALS_FILE", str(tmp_path / "credentials"))
+    monkeypatch.setenv("AWS_CONFIG_FILE", str(tmp_path / "config"))
+    monkeypatch.setenv("AWS_EC2_METADATA_DISABLED", "true")
+    for env_var in (
+        "AWS_PROFILE",
+        "AWS_DEFAULT_PROFILE",
+        "AWS_BEARER_TOKEN_BEDROCK",
+        "AWS_REGION_NAME",
+        "AWS_DEFAULT_REGION",
+    ):
+        monkeypatch.delenv(env_var, raising=False)
+
+
 class FakePayloadPart:
     def __init__(self, bytes_):
         self.bytes_ = bytes_

tests/unit/llms/bedrock/rerank/test_bedrock_rerank_header_forwarding.py

@@ -15,6 +15,21 @@ from litellm.llms.bedrock.base_aws_llm import Boto3CredentialsInfo
 from litellm.llms.bedrock.rerank.handler import BedrockRerankHandler
 from litellm.llms.custom_httpx.http_handler import AsyncHTTPHandler, HTTPHandler
 
+
+@pytest.fixture(autouse=True)
+def _isolate_host_aws_config(monkeypatch, tmp_path):
+    monkeypatch.setenv("AWS_SHARED_CREDENTIALS_FILE", str(tmp_path / "credentials"))
+    monkeypatch.setenv("AWS_CONFIG_FILE", str(tmp_path / "config"))
+    monkeypatch.setenv("AWS_EC2_METADATA_DISABLED", "true")
+    for env_var in (
+        "AWS_PROFILE",
+        "AWS_DEFAULT_PROFILE",
+        "AWS_BEARER_TOKEN_BEDROCK",
+        "AWS_REGION_NAME",
+        "AWS_DEFAULT_REGION",
+    ):
+        monkeypatch.delenv(env_var, raising=False)
+
 # Mock response for Bedrock rerank
 # Format based on Bedrock rerank API response structure
 bedrock_rerank_response = {
@@ -30,7 +45,8 @@ bedrock_rerank_response = {
 test_query = "What is the capital of the United States?"
 test_documents = [
     "Carson City is the capital city of the American state of Nevada.",
-    "The Commonwealth of the Northern Mariana Islands is a group of islands in the Pacific Ocean. Its capital is Saipan.",
+    "The Commonwealth of the Northern Mariana Islands is a group of islands in the Pacific Ocean. "
+    "Its capital is Saipan.",
     "Washington, D.C. is the capital of the United States.",
 ]
 

tests/unit/llms/bedrock/vector_stores/test_bedrock_vector_store_transformation.py

@@ -46,7 +46,8 @@ def test_transform_search_request_encodes_vector_store_id():
 
     assert (
         url
-        == "https://bedrock-agent-runtime.us-west-2.amazonaws.com/knowledgebases/..%2F..%2Fknowledgebases%2Fother%3Fx%3D1%23frag/retrieve"
+        == "https://bedrock-agent-runtime.us-west-2.amazonaws.com/knowledgebases/..%2F..%2Fknowledgebases%2Fother"
+        "%3Fx%3D1%23frag/retrieve"
     )
     assert body["retrievalQuery"].get("text") == "hello"
 

tests/unit/llms/bedrock_mantle/passthrough/test_bedrock_mantle_passthrough_transformation.py

@@ -109,7 +109,13 @@ def test_region_falls_back_to_the_mantle_default_without_any_hint(no_ambient_aws
         ({}, {"AWS_BEARER_TOKEN_BEDROCK": "aws-env-key"}, "aws-env-key"),
     ],
 )
-def test_sign_request_uses_the_deployment_bearer_token(no_ambient_aws, monkeypatch, litellm_params, env, expected_bearer):
+def test_sign_request_uses_the_deployment_bearer_token(
+    no_ambient_aws,
+    monkeypatch,
+    litellm_params,
+    env,
+    expected_bearer,
+):
     for name, value in env.items():
         monkeypatch.setenv(name, value)
     headers, body = BedrockMantlePassthroughConfig().sign_request(

tests/unit/llms/bytez/chat/test_bytez_chat_transformation.py

@@ -8,6 +8,32 @@ from litellm.llms.bytez.chat.transformation import BytezChatConfig, API_BASE, ve
 TEST_API_KEY = "MOCK_BYTEZ_API_KEY"
 TEST_MODEL_NAME = "google/gemma-3-4b-it"
 TEST_MODEL = f"bytez/{TEST_MODEL_NAME}"
+CAT_IMAGE_URL = (
+    "https://images.squarespace-cdn.com/content/v1/5452d441e4b0c188b51fef1a/1615326541809-TW01PVTOJ4PXQUX"
+    "VRLHI/male-orange-tabby-cat.jpg"
+)
+KAGGLE_AUDIO_URL = (
+    "https://storage.googleapis.com/kagglesdsdata/datasets/1736753/2838478/dataset/dataset/B_ANI01_MC_FN_"
+    "SIM01_101.wav?X-Goog-Algorithm=GOOG4-RSA-SHA256&X-Goog-Credential=databundle-worker-v2%40kaggle-1616"
+    "07.iam.gserviceaccount.com%2F20250711%2Fauto%2Fstorage%2Fgoog4_request&X-Goog-Date=20250711T192905Z&"
+    "X-Goog-Expires=345600&X-Goog-SignedHeaders=host&X-Goog-Signature=812b4bd6fcf9296f8e34f67664d900a81cf"
+    "81a4c8a4f439ce12befc89b4bef07c2645cab20ce5ba8f6b311dffa85aa05b70b4efbe53bced50a43a5e7622ea1ee0d8cc39"
+    "0679cdc6a6aae2c27f75debc1ce2361c595b3c9e1b8c88e2756ffc6b4f290af7f3dfa7232dc69ccc9a2181be756e0d538250"
+    "f9761a8b05ba1ac6c6b5d946f97a16aa14a5609ae62a2c4713c2077fcd34d129dbcdac6bb543ae547507b1a424e4fd09f817"
+    "000943c11507e0a74c514ec212b17427b7fc9e2ce87a250db1258645e4862a4261e3790fd99c9186148ad0653acd2b6a9468"
+    "adbeb94f17b5a685551037fd2cc9fe72fa405a006c0bd42d03be1e4c0dc4023ed3a77171edff3"
+)
+KAGGLE_VIDEO_URL = (
+    "https://storage.googleapis.com/kagglesdsdata/datasets/3957252/6888743/dog1.mp4?X-Goog-Algorithm=GOOG"
+    "4-RSA-SHA256&X-Goog-Credential=databundle-worker-v2%40kaggle-161607.iam.gserviceaccount.com%2F202507"
+    "11%2Fauto%2Fstorage%2Fgoog4_request&X-Goog-Date=20250711T193025Z&X-Goog-Expires=345600&X-Goog-Signed"
+    "Headers=host&X-Goog-Signature=810961d9abcbc2437954fdf19ef216deb65d3977eb354ec10af0d4644627cc6b143a5f"
+    "c6450996bae1787c09d26334de7cd6ff887510a5ac2a6eed3cfcc6673a47686c84c1f2b0bf543009388d83f2cd9551ad5f72"
+    "084513c6a7acd2c718849a4ebe951ccc5631bed014b0d115225c048b9f5de68673a37db24a98ad39cf3d0ba16fb764bf38eb"
+    "90c78c295c21a4ddac08c3c661b65efd511ccb86bacb87a2e2a97a06f53ea1c64d5dcf274001a61bc20867802549601301d9"
+    "99f5a5b2e49fd444b7db860c68e1c67df6e8edd5ad97171eaafb4fa1462453924ea4d78733be411cb6b5c910d4f829cd7189"
+    "c28dc1b22c8ae2a4da844a0d202e9e64bc7fb17947"
+)
 TEST_MESSAGES = [{"role": "user", "content": "Hello"}]
 
 
@@ -148,7 +174,7 @@ class TestBytezChatConfig:
                             "What color is this cat?",
                             {
                                 "type": "image_url",
-                                "url": "https://images.squarespace-cdn.com/content/v1/5452d441e4b0c188b51fef1a/1615326541809-TW01PVTOJ4PXQUXVRLHI/male-orange-tabby-cat.jpg",
+                                "url": CAT_IMAGE_URL,
                             },
                         ],
                     }
@@ -160,7 +186,7 @@ class TestBytezChatConfig:
                             {"type": "text", "text": "What color is this cat?"},
                             {
                                 "type": "image",
-                                "url": "https://images.squarespace-cdn.com/content/v1/5452d441e4b0c188b51fef1a/1615326541809-TW01PVTOJ4PXQUXVRLHI/male-orange-tabby-cat.jpg",
+                                "url": CAT_IMAGE_URL,
                             },
                         ],
                     }
@@ -174,7 +200,7 @@ class TestBytezChatConfig:
                             {"type": "text", "text": "What color is this cat?"},
                             {
                                 "type": "image_url",
-                                "url": "https://images.squarespace-cdn.com/content/v1/5452d441e4b0c188b51fef1a/1615326541809-TW01PVTOJ4PXQUXVRLHI/male-orange-tabby-cat.jpg",
+                                "url": CAT_IMAGE_URL,
                             },
                         ],
                     }
@@ -186,7 +212,7 @@ class TestBytezChatConfig:
                             {"type": "text", "text": "What color is this cat?"},
                             {
                                 "type": "image",
-                                "url": "https://images.squarespace-cdn.com/content/v1/5452d441e4b0c188b51fef1a/1615326541809-TW01PVTOJ4PXQUXVRLHI/male-orange-tabby-cat.jpg",
+                                "url": CAT_IMAGE_URL,
                             },
                         ],
                     }
@@ -200,7 +226,7 @@ class TestBytezChatConfig:
                             {"type": "text", "text": "What kind of cat meow is this?"},
                             {
                                 "type": "input_audio",
-                                "url": "https://storage.googleapis.com/kagglesdsdata/datasets/1736753/2838478/dataset/dataset/B_ANI01_MC_FN_SIM01_101.wav?X-Goog-Algorithm=GOOG4-RSA-SHA256&X-Goog-Credential=databundle-worker-v2%40kaggle-161607.iam.gserviceaccount.com%2F20250711%2Fauto%2Fstorage%2Fgoog4_request&X-Goog-Date=20250711T192905Z&X-Goog-Expires=345600&X-Goog-SignedHeaders=host&X-Goog-Signature=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",
+                                "url": KAGGLE_AUDIO_URL,
                             },
                         ],
                     }
@@ -212,7 +238,7 @@ class TestBytezChatConfig:
                             {"type": "text", "text": "What kind of cat meow is this?"},
                             {
                                 "type": "audio",
-                                "url": "https://storage.googleapis.com/kagglesdsdata/datasets/1736753/2838478/dataset/dataset/B_ANI01_MC_FN_SIM01_101.wav?X-Goog-Algorithm=GOOG4-RSA-SHA256&X-Goog-Credential=databundle-worker-v2%40kaggle-161607.iam.gserviceaccount.com%2F20250711%2Fauto%2Fstorage%2Fgoog4_request&X-Goog-Date=20250711T192905Z&X-Goog-Expires=345600&X-Goog-SignedHeaders=host&X-Goog-Signature=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",
+                                "url": KAGGLE_AUDIO_URL,
                             },
                         ],
                     }
@@ -226,7 +252,7 @@ class TestBytezChatConfig:
                             {"type": "text", "text": "What kind of dog is this?"},
                             {
                                 "type": "video_url",
-                                "url": "https://storage.googleapis.com/kagglesdsdata/datasets/3957252/6888743/dog1.mp4?X-Goog-Algorithm=GOOG4-RSA-SHA256&X-Goog-Credential=databundle-worker-v2%40kaggle-161607.iam.gserviceaccount.com%2F20250711%2Fauto%2Fstorage%2Fgoog4_request&X-Goog-Date=20250711T193025Z&X-Goog-Expires=345600&X-Goog-SignedHeaders=host&X-Goog-Signature=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",
+                                "url": KAGGLE_VIDEO_URL,
                             },
                         ],
                     }
@@ -238,7 +264,7 @@ class TestBytezChatConfig:
                             {"type": "text", "text": "What kind of dog is this?"},
                             {
                                 "type": "video",
-                                "url": "https://storage.googleapis.com/kagglesdsdata/datasets/3957252/6888743/dog1.mp4?X-Goog-Algorithm=GOOG4-RSA-SHA256&X-Goog-Credential=databundle-worker-v2%40kaggle-161607.iam.gserviceaccount.com%2F20250711%2Fauto%2Fstorage%2Fgoog4_request&X-Goog-Date=20250711T193025Z&X-Goog-Expires=345600&X-Goog-SignedHeaders=host&X-Goog-Signature=810961d9abcbc2437954fdf19ef216deb65d3977eb354ec10af0d4644627cc6b143a5fc6450996bae1787c09d26334de7cd6ff887510a5ac2a6eed3cfcc6673a47686c84c1f2b0bf543009388d83f2cd9551ad5f72084513c6a7acd2c718849a4ebe951ccc5631bed014b0d115225c048b9f5de68673a37db24a98ad39cf3d0ba16fb764bf38eb90c78c295c21a4ddac08c3c661b65efd511ccb86bacb87a2e2a97a06f53ea1c64d5dcf274001a61bc20867802549601301d999f5a5b2e49fd444b7db860c68e1c67df6e8edd5ad97171eaafb4fa1462453924ea4d78733be411cb6b5c910d4f829cd7189c28dc1b22c8ae2a4da844a0d202e9e64bc7fb17947",
+                                "url": KAGGLE_VIDEO_URL,
                             },
                         ],
                     }

tests/unit/llms/chat/test_converse_handler.py

@@ -106,7 +106,10 @@ class TestBedrockRegionInModelPath:
         ), f"modelId mismatch for {model!r}: got {model_id!r}, expected {expected_model_id!r}"
         assert (
             optional_params.get("aws_region_name") == expected_region
-        ), f"region mismatch for {model!r}: got {optional_params.get('aws_region_name')!r}, expected {expected_region!r}"
+        ), (
+            f"region mismatch for {model!r}: "
+            f"got {optional_params.get('aws_region_name')!r}, expected {expected_region!r}"
+        )
 
     def test_explicit_aws_region_name_not_overridden(self):
         """