chore: merge main into the OpenAI SDK 3 branch

The only conflicts were test imports. uv.lock differs from main only in the openai chain

enterprise/litellm_enterprise/proxy/common_utils/check_responses_cost.py

@@ -30,7 +30,7 @@ from collections.abc import Awaitable, Callable, Mapping
 from dataclasses import dataclass
 from datetime import datetime, timedelta, timezone
 from types import MappingProxyType
-from typing import TYPE_CHECKING, Final, cast
+from typing import TYPE_CHECKING, Final
 
 from litellm_enterprise.proxy.common_utils.managed_object_attribution import (
     ManagedObjectCreatorAttribution,
@@ -290,12 +290,14 @@ class CheckResponsesCost:
         sees provider env vars, so it fails for every deployment whose credentials
         live in the config; the row then never leaves ``queued``.
         """
-        if not via_router:
-            return await litellm.aget_responses(response_id=response_id, litellm_metadata=litellm_metadata)
-        router_response = await self.llm_router.aget_responses(
-            response_id=response_id, litellm_metadata=litellm_metadata
+        retrieved: Final[object] = (
+            await self.llm_router.aget_responses(response_id=response_id, litellm_metadata=litellm_metadata)
+            if via_router
+            else await litellm.aget_responses(response_id=response_id, litellm_metadata=litellm_metadata)
         )
-        return cast(ResponsesAPIResponse, router_response)
+        if not isinstance(retrieved, ResponsesAPIResponse):
+            raise TypeError(f"Retrieving {response_id} did not return a response object")
+        return retrieved
 
     async def _fetch_unbilled(self, target: _PollTarget) -> ResponsesAPIResponse:
         return await self._get_response(

litellm/__init__.py

@@ -39,6 +39,7 @@ from litellm._logging import (
     turn_on_json,
     log_level,
 )
+from litellm._logging import suppress_startup_output as suppress_startup_output
 import re
 from litellm.constants import (
     DEFAULT_BATCH_SIZE,
@@ -684,6 +685,7 @@ moonshot_models: Set = set()
 publicai_models: Set = set()
 darkbloom_models: Set = set()
 openinfer_models: Final[Set[str]] = set()
+kyma_models: Final[set[str]] = set()  # mutable-ok: populated when the shared model catalog loads
 streamlake_models: Final[Set[str]] = set()
 zerogpu_models: Final[Set[str]] = set()
 v0_models: Set = set()
@@ -980,6 +982,8 @@ def _populate_provider_model_sets(model_cost_map: Mapping[str, Mapping[str, obje
             darkbloom_models.add(key)
         elif value.get("litellm_provider") == "openinfer":
             openinfer_models.add(key)
+        elif value.get("litellm_provider") == "kyma":
+            kyma_models.add(key)
         elif value.get("litellm_provider") == "streamlake":
             streamlake_models.add(key)
         elif value.get("litellm_provider") == "zerogpu":
@@ -1158,6 +1162,7 @@ model_list = list(
     | publicai_models
     | darkbloom_models
     | openinfer_models
+    | kyma_models
     | streamlake_models
     | zerogpu_models
     | v0_models
@@ -1289,6 +1294,7 @@ def _build_models_by_provider() -> dict:
         "publicai": publicai_models,
         "darkbloom": darkbloom_models,
         "openinfer": openinfer_models,
+        "kyma": kyma_models,
         "streamlake": streamlake_models,
         "zerogpu": zerogpu_models,
         "v0": v0_models,

litellm/_internal_context.py

@@ -36,6 +36,8 @@ REDIS_FAMILIES_METADATA_KEY: Final = "families"
 
 _service_caller: Final[ContextVar[str | None]] = ContextVar("service_caller", default=None)
 
+_mirrored_call: Final[ContextVar[bool]] = ContextVar("mirrored_call", default=False)
+
 
 @contextmanager
 def internal_sub_call() -> Generator[None]:
@@ -123,6 +125,30 @@ def current_service_caller() -> str | None:
     return _service_caller.get()
 
 
+@contextmanager
+def mirrored_call() -> Generator[None]:
+    """A silent_model mirror of a request: every litellm call inside, its retries and fallbacks included."""
+    token: Final = _mirrored_call.set(True)
+    try:
+        yield
+    finally:
+        _restore(_mirrored_call, token)
+
+
+def in_mirrored_call() -> bool:
+    return _mirrored_call.get()
+
+
+@contextmanager
+def outside_mirrored_call() -> Generator[None]:
+    """Logging a mirrored call: the calls its callbacks start are not mirrors themselves."""
+    token: Final = _mirrored_call.set(False)
+    try:
+        yield
+    finally:
+        _restore(_mirrored_call, token)
+
+
 @contextmanager
 def pinned_billing_time(moment: datetime) -> Generator[None]:
     """Price every rate lookup inside this block at ``moment`` rather than at each one's own clock read."""

litellm/_logging.py

@@ -10,10 +10,12 @@ import sys
 from collections.abc import Iterator, Mapping
 from datetime import datetime
 from logging import Formatter
+from types import FrameType
 from typing import Final, TextIO
 from urllib.parse import unquote
 
 from pydantic import TypeAdapter
+from typing_extensions import ReadOnly, TypedDict
 
 import litellm
 from litellm.constants import (
@@ -598,6 +600,7 @@ _LOG_FORMAT_PREFIX: Final = "%(asctime)s - %(name)s:%(levelname)s"
 _LOG_FORMAT_SUFFIX: Final = ": %(filename)s:%(lineno)s - %(message)s"
 _PLAIN_LOG_FORMAT: Final = _LOG_FORMAT_PREFIX + _LOG_FORMAT_SUFFIX
 _COLOR_LOG_FORMAT: Final = f"\033[92m{_LOG_FORMAT_PREFIX}\033[0m{_LOG_FORMAT_SUFFIX}"
+_ANSI_ESCAPE: Final = re.compile(r"\x1b\[[0-9;]*m")
 
 
 def _stream_is_tty(stream: TextIO | None) -> bool:
@@ -612,15 +615,28 @@ def _stream_is_tty(stream: TextIO | None) -> bool:
         return False
 
 
-def _plain_log_format(stdout: TextIO | None, stderr: TextIO | None) -> str:
-    """The plain-text log format, colorized only when both streams are an interactive terminal.
+def _color_enabled(stream: TextIO | None) -> bool:
+    """True when ANSI colors written to the stream reach an interactive terminal.
 
     Honors the NO_COLOR convention from no-color.org: color is disabled when
     NO_COLOR is present with a non-empty value.
     """
-    if os.environ.get("NO_COLOR"):
-        return _PLAIN_LOG_FORMAT
-    return _COLOR_LOG_FORMAT if _stream_is_tty(stdout) and _stream_is_tty(stderr) else _PLAIN_LOG_FORMAT
+    return not os.environ.get("NO_COLOR") and _stream_is_tty(stream)
+
+
+def _terminal_colors(stdout: TextIO | None, stderr: TextIO | None) -> bool:
+    """True when both streams are an interactive color terminal, the one rule for log lines and startup prints."""
+    return _color_enabled(stdout) and _color_enabled(stderr)
+
+
+def terminal_text(message: str) -> str:
+    """The message with its ANSI colors kept only on a color terminal, by the rule the plain log format uses."""
+    return message if _terminal_colors(sys.stdout, sys.stderr) else _ANSI_ESCAPE.sub("", message)
+
+
+def _plain_log_format(stdout: TextIO | None, stderr: TextIO | None) -> str:
+    """The plain-text log format, colorized only when both streams are an interactive terminal."""
+    return _COLOR_LOG_FORMAT if _terminal_colors(stdout, stderr) else _PLAIN_LOG_FORMAT
 
 
 class LevelRoutingStreamHandler(logging.StreamHandler):
@@ -642,8 +658,8 @@ class LevelRoutingStreamHandler(logging.StreamHandler):
         super().emit(record)
 
 
-def _parse_json_logs_env(value: str | None) -> bool:
-    """Strict opt-in parse for the JSON_LOGS env var: only "true" (any case) enables JSON logs.
+def _parse_env_flag(value: str | None) -> bool:
+    """Strict opt-in parse for a boolean env var such as JSON_LOGS: only "true" (any case) enables it.
 
     Matches the reader in litellm-proxy-extras/_logging.py. The previous
     bool(os.getenv(...)) treated any non-empty value, including "false" and "0",
@@ -656,7 +672,18 @@ def resolve_log_level(log_level: str) -> int:
     return getattr(logging, log_level.upper())
 
 
-json_logs: Final = _parse_json_logs_env(os.getenv("JSON_LOGS"))
+json_logs: Final = _parse_env_flag(os.getenv("JSON_LOGS"))
+suppress_startup_output: Final = _parse_env_flag(os.getenv("LITELLM_SUPPRESS_STARTUP_OUTPUT"))
+
+
+class StartupOutputSettings(TypedDict, total=False):
+    json_logs: ReadOnly[object]
+    suppress_startup_output: ReadOnly[object]
+
+
+STARTUP_OUTPUT_SETTINGS: Final = TypeAdapter(StartupOutputSettings)
+
+
 # Create a handler for the logger (you may need to adapt this based on your needs)
 log_level: Final = os.getenv("LITELLM_LOG", "DEBUG")
 numeric_level: Final[int] = resolve_log_level(log_level)
@@ -743,7 +770,7 @@ class JsonFormatter(Formatter):
         return dt.isoformat()
 
     def format(self, record):
-        message_str: Final = record.getMessage()
+        message_str: Final = _ANSI_ESCAPE.sub("", record.getMessage())
         json_record: Final[dict[str, object]] = {
             "message": message_str,
             "level": record.levelname,
@@ -799,12 +826,14 @@ class CorrelationPlainFormatter(logging.Formatter):
     """Appends trace_id/session_id to plain-text log lines stamped by CorrelationContextFilter.
 
     Mirrors JsonFormatter's handling of these two fields so request_correlation_in_logs
-    behaves the same whether or not json_logs is enabled.
+    behaves the same whether or not json_logs is enabled. Unless it was built with the color format,
+    it also drops the ANSI color codes a message carries, which a piped log would store as text.
     """
 
     def format(self, record: logging.LogRecord) -> str:
         rendered: Final = super().format(record)
-        formatted: Final = rendered if _is_redacted(record) else redact_string(rendered)
+        uncolored: Final = rendered if self._fmt == _COLOR_LOG_FORMAT else _ANSI_ESCAPE.sub("", rendered)
+        formatted: Final = uncolored if _is_redacted(record) else redact_string(uncolored)
         trace_id: Final = getattr(record, "trace_id", None)
         session_id: Final = getattr(record, "session_id", None)
         if not trace_id and not session_id:
@@ -898,29 +927,46 @@ verbose_proxy_logger.addFilter(_diagnostic_filter)
 verbose_proxy_stdout_logger.addFilter(_diagnostic_filter)
 verbose_logger.addFilter(_diagnostic_filter)
 
-_ANSI_ESCAPE: Final = re.compile(r"\x1b\[[0-9;]*m")
-
 
 def print_startup_message(message: str) -> None:
     """Write a boot-time message meant for the operator's terminal.
 
-    Plain mode prints it unchanged, colors and spacing included. With JSON logs on, every line of
-    stdout is read by a log pipeline, so the message goes out as one INFO record of the proxy logger
-    instead, on stdout like the print whatever handlers the logging config gave that logger, with the
-    ANSI color codes and surrounding blank lines removed, located at the caller and emitted at any log
-    level, whole and unredacted as the print showed it; a blank spacer produces no record.
+    LITELLM_SUPPRESS_STARTUP_OUTPUT or litellm_settings.suppress_startup_output turns every one of them
+    off; a warning goes through print_startup_warning instead. Plain mode prints it with its spacing,
+    keeping the ANSI colors only on a color terminal by the same check as the plain log format, so a
+    container log never stores escape codes.
+    With JSON logs on, every line of stdout is read by a log pipeline, so the message goes out as one
+    INFO record of the proxy logger instead, on stdout like the print whatever handlers the logging
+    config gave that logger, with the ANSI color codes and surrounding blank lines removed, located at
+    the caller and emitted at any log level, whole and unredacted as the print showed it; a blank
+    spacer produces no record.
     """
+    if litellm.suppress_startup_output is True:
+        return
+    frame: Final = inspect.currentframe()
+    _write_startup_message(message, frame.f_back if frame is not None else None, logging.INFO)
+
+
+def print_startup_warning(message: str) -> None:
+    """Write a boot-time warning as print_startup_message writes a message, even with suppress_startup_output
+    on, since it names something the operator has to fix. With JSON logs on it is a WARNING record, on
+    stdout where the print put it, so a collector that filters on the level field keeps it."""
+    frame: Final = inspect.currentframe()
+    _write_startup_message(message, frame.f_back if frame is not None else None, logging.WARNING)
+
+
+def _write_startup_message(message: str, caller: FrameType | None, level: int) -> None:
     if not litellm.json_logs:
-        print(message)  # noqa: T201  # plain mode: this is the operator-terminal print it replaces
+        import click
+
+        click.echo(terminal_text(message))
         return
     text: Final = _ANSI_ESCAPE.sub("", message).strip("\n")
     if not text:
         return
-    frame: Final = inspect.currentframe()
-    caller: Final = frame.f_back if frame is not None else None
     record: Final = verbose_proxy_logger.makeRecord(
         verbose_proxy_logger.name,
-        logging.INFO,
+        level,
         caller.f_code.co_filename if caller is not None else __file__,
         caller.f_lineno if caller is not None else 0,
         text,
@@ -934,8 +980,8 @@ def print_startup_message(message: str) -> None:
 
 
 def _startup_json_handler() -> logging.Handler:
-    """The proxy's JSON handler on stdout, filters included, for one startup message."""
-    handler: Final = LevelRoutingStreamHandler()
+    """The proxy's JSON handler on stdout, filters included, for one startup message or warning."""
+    handler: Final = logging.StreamHandler(sys.stdout)
     handler.setFormatter(JsonFormatter())
     handler.addFilter(_secret_filter)
     handler.addFilter(_correlation_filter)

litellm/anthropic_beta_gated_fields.py

@@ -0,0 +1,325 @@
+"""Body fields that an ``anthropic-beta`` value gates.
+
+``anthropic_beta_headers_config.json`` says which beta each provider leg
+carries. When a leg drops a beta, the request fields that beta unlocks
+would make the provider 400. Under ``drop_params`` the same request degrades on
+that leg by dropping those fields too; without it the request goes out as sent
+and the provider answers its own 400, like any unsupported param. Every helper
+here is copy-on-write: the caller's containers are returned untouched when
+nothing is stripped, and a shallow copy is returned otherwise, so a fallback
+leg still sends the full request.
+"""
+
+from collections.abc import Mapping, Sequence, Set
+from dataclasses import dataclass
+from types import MappingProxyType
+from typing import Final
+
+from pydantic import TypeAdapter, ValidationError
+
+from litellm._logging import verbose_logger
+from litellm.litellm_core_utils.core_helpers import drops_unsupported_params
+from litellm.types.llms.anthropic import (
+    ANTHROPIC_BETA_HEADER_VALUES,
+    ANTHROPIC_MID_CONVERSATION_OUTPUT_CONFIG_BETA_HEADER,
+    ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER,
+    ANTHROPIC_THINKING_BINDING_CONTROLS_BETA_HEADER,
+    ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER,
+)
+
+ANTHROPIC_MID_CONVERSATION_SYSTEM_CLEAR_AT_BETA_HEADER: Final = "mid-conversation-system-clear-at-2026-08-21"
+ANTHROPIC_INLINE_TOOLS_BETA_HEADER: Final = "inline-tools-2026-09-15"
+
+THINKING_BLOCK_BINDING: Final = "thinking.block_binding"
+THINKING_DISPLAY_UPDATES: Final = "thinking.display=updates"
+SYSTEM_CLEAR_AT: Final = "messages[].system.clear_at"
+SYSTEM_OUTPUT_CONFIG: Final = "messages[].system.output_config"
+SYSTEM_TOOL_CHANGES: Final = "messages[].system.content[].tool_addition|tool_removal"
+SYSTEM_MESSAGE_EMPTIED: Final = "messages[].system (emptied by the strip)"
+
+_TOOL_CHANGE_BLOCK_TYPES: Final = frozenset({"tool_addition", "tool_removal"})
+
+BETA_GATED_FIELDS: Final[Mapping[str, tuple[str, ...]]] = MappingProxyType(
+    {
+        ANTHROPIC_THINKING_BINDING_CONTROLS_BETA_HEADER: (THINKING_BLOCK_BINDING,),
+        ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER: (THINKING_DISPLAY_UPDATES,),
+        ANTHROPIC_MID_CONVERSATION_SYSTEM_CLEAR_AT_BETA_HEADER: (SYSTEM_CLEAR_AT,),
+        ANTHROPIC_MID_CONVERSATION_OUTPUT_CONFIG_BETA_HEADER: (SYSTEM_OUTPUT_CONFIG,),
+        ANTHROPIC_BETA_HEADER_VALUES.PER_TURN_CONTROL_2026_07_01.value: (SYSTEM_OUTPUT_CONFIG,),
+        ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER: (SYSTEM_TOOL_CHANGES,),
+        ANTHROPIC_INLINE_TOOLS_BETA_HEADER: (SYSTEM_TOOL_CHANGES,),
+    }
+)
+
+
+def _betas_unlocking(field: str) -> frozenset[str]:
+    return frozenset(beta for beta, fields in BETA_GATED_FIELDS.items() if field in fields)
+
+
+_FIELD_BETAS: Final[Mapping[str, frozenset[str]]] = MappingProxyType(
+    {
+        field: _betas_unlocking(field) for fields in BETA_GATED_FIELDS.values() for field in fields
+    }  # comprehension-ok: flattening the beta-to-fields map into a field-to-betas index
+)
+
+
+def field_is_gated(field: str, active_betas: Set[str]) -> bool:
+    """True when no active beta unlocks ``field``, so the field must be stripped."""
+    return _FIELD_BETAS[field].isdisjoint(active_betas)
+
+
+_MAPPING_ADAPTER: Final = TypeAdapter(Mapping[str, object])
+_SEQUENCE_ADAPTER: Final = TypeAdapter(tuple[object, ...])
+_HEADERS_ADAPTER: Final = TypeAdapter(Mapping[str, str])
+
+
+def _as_mapping(value: object) -> Mapping[str, object] | None:
+    try:
+        return _MAPPING_ADAPTER.validate_python(value, strict=True)
+    except ValidationError:
+        return None
+
+
+def _as_sequence(value: object) -> tuple[object, ...] | None:
+    try:
+        return _SEQUENCE_ADAPTER.validate_python(value)
+    except ValidationError:
+        return None
+
+
+def _is_tool_change_block(block: object) -> bool:
+    mapping: Final = _as_mapping(block)
+    block_type: Final = mapping.get("type") if mapping is not None else None
+    return isinstance(block_type, str) and block_type in _TOOL_CHANGE_BLOCK_TYPES
+
+
+def _header_beta_values(headers: Mapping[str, str]) -> frozenset[str]:
+    joined: Final = ",".join(value for key, value in headers.items() if key.lower() == "anthropic-beta")
+    return frozenset(piece.strip() for piece in joined.split(",") if piece.strip())
+
+
+def _validated_headers(headers: object) -> Mapping[str, str]:
+    try:
+        return _HEADERS_ADAPTER.validate_python(headers)
+    except ValidationError:
+        return MappingProxyType({})
+
+
+def active_beta_values(headers: object, body_betas: object = ()) -> frozenset[str]:
+    """The ``anthropic-beta`` values a request will send: the comma-joined header (key matched
+    case-insensitively) plus the body's ``anthropic_beta`` list on legs that carry betas in the body."""
+    body_values: Final = _as_sequence(body_betas) or ()
+    body_betas_as_strings: Final = frozenset(beta for beta in body_values if isinstance(beta, str))
+    return _header_beta_values(_validated_headers(headers)) | body_betas_as_strings
+
+
+@dataclass(frozen=True, slots=True)
+class StrippedThinking:
+    thinking: object
+    stripped: tuple[str, ...]
+
+
+@dataclass(frozen=True, slots=True)
+class StrippedMessages:
+    messages: Sequence[object]
+    stripped: tuple[str, ...]
+
+
+@dataclass(frozen=True, slots=True)
+class StrippedRequest:
+    request: Mapping[str, object]
+    stripped: tuple[str, ...]
+
+
+def strip_beta_gated_thinking(thinking: object, active_betas: Set[str], *, drop_params: object) -> StrippedThinking:
+    """Drop ``thinking`` sub-fields whose beta is absent. Returns the same object when nothing is gated."""
+    mapping: Final = _as_mapping(thinking)
+    if mapping is None or not drops_unsupported_params(drop_params):
+        return StrippedThinking(thinking=thinking, stripped=())
+    binding_gated: Final = "block_binding" in mapping and field_is_gated(THINKING_BLOCK_BINDING, active_betas)
+    display_gated: Final = mapping.get("display") == "updates" and field_is_gated(
+        THINKING_DISPLAY_UPDATES, active_betas
+    )
+    if not binding_gated and not display_gated:
+        return StrippedThinking(thinking=thinking, stripped=())
+    gated_keys: Final = frozenset(
+        {*(("block_binding",) if binding_gated else ()), *(("display",) if display_gated else ())}
+    )
+    stripped: Final = (
+        *((THINKING_BLOCK_BINDING,) if binding_gated else ()),
+        *((THINKING_DISPLAY_UPDATES,) if display_gated else ()),
+    )
+    return StrippedThinking(
+        thinking={key: value for key, value in mapping.items() if key not in gated_keys},
+        stripped=stripped,
+    )
+
+
+def _strip_beta_gated_system_message(message: Mapping[str, object], active_betas: Set[str]) -> StrippedMessages:
+    """One mid-conversation ``role: system`` message. ``messages`` holds the rewritten message, or nothing
+    when the strip left it with neither content nor ``output_config``, since such a system turn is itself a 400."""
+    clear_at_gated: Final = "clear_at" in message and field_is_gated(SYSTEM_CLEAR_AT, active_betas)
+    output_config_gated: Final = "output_config" in message and field_is_gated(SYSTEM_OUTPUT_CONFIG, active_betas)
+    content: Final = message.get("content")
+    blocks: Final = _as_sequence(content) if not isinstance(content, str) else None
+    tool_changes_gated: Final = (
+        blocks is not None
+        and any(_is_tool_change_block(block) for block in blocks)
+        and field_is_gated(SYSTEM_TOOL_CHANGES, active_betas)
+    )
+    if not clear_at_gated and not output_config_gated and not tool_changes_gated:
+        return StrippedMessages(messages=(message,), stripped=())
+    kept_content: Final[object] = (
+        [block for block in blocks if not _is_tool_change_block(block)]
+        if tool_changes_gated and blocks is not None
+        else content
+    )
+    gated_keys: Final = frozenset(
+        {*(("clear_at",) if clear_at_gated else ()), *(("output_config",) if output_config_gated else ())}
+    )
+    rewritten: Final = {
+        key: (kept_content if key == "content" else value) for key, value in message.items() if key not in gated_keys
+    }
+    stripped: Final = (
+        *((SYSTEM_CLEAR_AT,) if clear_at_gated else ()),
+        *((SYSTEM_OUTPUT_CONFIG,) if output_config_gated else ()),
+        *((SYSTEM_TOOL_CHANGES,) if tool_changes_gated else ()),
+    )
+    if kept_content in (None, "", []) and "output_config" not in rewritten:
+        return StrippedMessages(messages=(), stripped=(*stripped, SYSTEM_MESSAGE_EMPTIED))
+    return StrippedMessages(messages=(rewritten,), stripped=stripped)
+
+
+def _strip_beta_gated_message(message: object, active_betas: Set[str]) -> StrippedMessages:
+    mapping: Final = _as_mapping(message)
+    if mapping is None or mapping.get("role") != "system":
+        return StrippedMessages(messages=(message,), stripped=())
+    return _strip_beta_gated_system_message(mapping, active_betas)
+
+
+def _ends_on_assistant(messages: Sequence[object]) -> bool:
+    last: Final = _as_mapping(messages[-1]) if messages else None
+    return last is not None and last.get("role") == "assistant"
+
+
+def strip_beta_gated_system_messages(
+    messages: Sequence[object], active_betas: Set[str], *, drop_params: object
+) -> StrippedMessages:
+    """Drop the gated fields of every mid-conversation ``role: system`` message. Returns the same
+    sequence when nothing is gated, or when dropping an emptied system turn would leave the conversation
+    ending on an assistant turn, which the model would read as a prefill to continue."""
+    if not drops_unsupported_params(drop_params):
+        return StrippedMessages(messages=messages, stripped=())
+    per_message: Final = tuple(_strip_beta_gated_message(message, active_betas) for message in messages)
+    stripped: Final = tuple(
+        name for result in per_message for name in result.stripped
+    )  # comprehension-ok: flattening per-message strip reports
+    if not stripped:
+        return StrippedMessages(messages=messages, stripped=())
+    kept: Final = [
+        message for result in per_message for message in result.messages
+    ]  # comprehension-ok: flattening per-message results
+    if _ends_on_assistant(kept) and not _ends_on_assistant(messages):
+        return StrippedMessages(messages=messages, stripped=())
+    return StrippedMessages(messages=kept, stripped=stripped)
+
+
+def strip_beta_gated_request_fields(
+    request: Mapping[str, object], active_betas: Set[str], *, drop_params: object
+) -> StrippedRequest:
+    """Apply both strips to an Anthropic ``/v1/messages`` shaped body. Returns the same mapping when
+    nothing is gated, otherwise a shallow copy with ``thinking`` and ``messages`` replaced."""
+    thinking: Final = strip_beta_gated_thinking(request.get("thinking"), active_betas, drop_params=drop_params)
+    messages: Final = request.get("messages")
+    message_sequence: Final = _as_sequence(messages) if not isinstance(messages, str) else None
+    system_messages: Final = (
+        strip_beta_gated_system_messages(message_sequence, active_betas, drop_params=drop_params)
+        if message_sequence is not None
+        else StrippedMessages(messages=(), stripped=())
+    )
+    stripped: Final = (*thinking.stripped, *system_messages.stripped)
+    if not stripped:
+        return StrippedRequest(request=request, stripped=())
+    rewritten: Final = {
+        **request,
+        **({"thinking": thinking.thinking} if thinking.stripped else {}),
+        **({"messages": system_messages.messages} if system_messages.stripped else {}),
+    }
+    return StrippedRequest(request=rewritten, stripped=stripped)
+
+
+@dataclass(frozen=True, slots=True)
+class StrippedNativeRequest:
+    thinking: object
+    thinking_stripped: bool
+    messages: Sequence[object]
+    messages_stripped: bool
+    stripped: tuple[str, ...]
+
+
+def unstripped_native_request(optional_params: object, messages: object) -> StrippedNativeRequest:
+    """The identity result for legs that strip later, once their final ``anthropic_beta`` list exists."""
+    params: Final = _as_mapping(optional_params) or MappingProxyType({})
+    message_sequence: Final = _as_sequence(messages) if not isinstance(messages, str) else None
+    return StrippedNativeRequest(
+        thinking=params.get("thinking"),
+        thinking_stripped=False,
+        messages=message_sequence if message_sequence is not None else (),
+        messages_stripped=False,
+        stripped=(),
+    )
+
+
+def strip_beta_gated_native_request(
+    headers: object, optional_params: object, messages: object, *, body_betas: object = (), drop_params: object
+) -> StrippedNativeRequest:
+    """The ``/v1/messages`` seam, fed the final (already filtered) headers, the final body ``anthropic_beta``
+    list on legs that carry betas in the body, plus the body parts it may rewrite."""
+    if not drops_unsupported_params(drop_params):
+        return unstripped_native_request(optional_params, messages)
+    active_betas: Final = active_beta_values(headers, body_betas)
+    params: Final = _as_mapping(optional_params) or MappingProxyType({})
+    thinking: Final = strip_beta_gated_thinking(params.get("thinking"), active_betas, drop_params=drop_params)
+    message_sequence: Final = _as_sequence(messages) if not isinstance(messages, str) else None
+    system_messages: Final = (
+        strip_beta_gated_system_messages(message_sequence, active_betas, drop_params=drop_params)
+        if message_sequence is not None
+        else StrippedMessages(messages=(), stripped=())
+    )
+    return StrippedNativeRequest(
+        thinking=thinking.thinking,
+        thinking_stripped=bool(thinking.stripped),
+        messages=system_messages.messages,
+        messages_stripped=bool(system_messages.stripped),
+        stripped=(*thinking.stripped, *system_messages.stripped),
+    )
+
+
+def log_beta_gated_strips(stripped: Sequence[str], model: str | None, provider: str) -> None:
+    if not stripped:
+        return
+    described: Final = ", ".join(
+        f"{field} (unlocked by anthropic-beta {' or '.join(sorted(_FIELD_BETAS[field]))})"
+        if field in _FIELD_BETAS
+        else field
+        for field in sorted(set(stripped))
+    )
+    verbose_logger.warning(
+        "Dropped %s for model=%s provider=%s because drop_params is on and the provider would reject them on "
+        "this request, either since this provider leg does not send that anthropic-beta (see "
+        "anthropic_beta_headers_config.json) or since the model turns mid-conversation system messages into user "
+        "messages.",
+        described,
+        model,
+        provider,
+    )
+
+
+def strip_beta_gated_request_body(
+    request: Mapping[str, object], headers: object, *, drop_params: object, model: str | None, provider: str
+) -> Mapping[str, object]:
+    gated: Final = strip_beta_gated_request_fields(
+        request, active_beta_values(headers, request.get("anthropic_beta") or ()), drop_params=drop_params
+    )
+    log_beta_gated_strips(gated.stripped, model=model, provider=provider)
+    return gated.request

litellm/anthropic_beta_headers_config.json

@@ -248,6 +248,7 @@
     "text_editor_20241022": null,
     "text_editor_20250124": null,
     "thinking-binding-controls-2026-08-01": "thinking-binding-controls-2026-08-01",
+    "thinking-display-updates-2026-08-18": "thinking-display-updates-2026-08-18",
     "token-efficient-tools-2025-02-19": "token-efficient-tools-2025-02-19",
     "web-fetch-2025-09-10": null,
     "web-search-2025-03-05": "web-search-2025-03-05"

litellm/anthropic_beta_headers_manager.py

@@ -241,6 +241,13 @@ def get_provider_name(provider: str) -> str:
     return aliases.get(provider, provider)
 
 
+def has_beta_headers_config(provider: str) -> bool:
+    config: Final = _bundled_beta_headers_config()
+    aliases: Final = config.get("provider_aliases", {})
+    canonical_provider: Final = (aliases.get(provider) if isinstance(aliases, Mapping) else None) or provider
+    return canonical_provider in config
+
+
 def filter_and_transform_beta_headers(
     beta_headers: list[str],
     provider: str,
@@ -397,7 +404,7 @@ def update_request_with_filtered_beta(
     headers: dict,
     request_data: dict,
     provider: str,
-) -> tuple[dict, dict]:
+) -> tuple[dict[str, str], dict[str, object]]:
     """
     Update both headers and request body beta fields based on provider support.
     Modifies both dicts in place and returns them.

litellm/caching/caching.py

@@ -455,17 +455,33 @@ class Cache:
         3. Else use the `model` passed in kwargs, plus `model_id` when set (e.g. a SageMaker inference component)
         """
         litellm_params: Final[dict] = kwargs.get("litellm_params", {}) or {}
-        metadata_sources: Final[tuple[dict, ...]] = (
-            kwargs.get("metadata") or {},
-            kwargs.get("litellm_metadata") or {},
-            litellm_params.get("metadata") or {},
-            litellm_params.get("litellm_metadata") or {},
+        metadata_sources: Final = cast(  # cast-ok: each source is the untyped request's metadata dict or empty
+            "tuple[dict[str, object], dict[str, object], dict[str, object], dict[str, object]]",
+            (
+                kwargs.get("metadata") or {},
+                kwargs.get("litellm_metadata") or {},
+                litellm_params.get("metadata") or {},
+                litellm_params.get("litellm_metadata") or {},
+            ),
         )
-        model_group: Final[str | None] = next(
-            (source["model_group"] for source in metadata_sources if source.get("model_group")), None
+        metadata, litellm_metadata, params_metadata, params_litellm_metadata = metadata_sources
+        call_model_info: Final[object] = kwargs.get("model_info")
+        routed_source: Final = next(
+            (
+                source
+                for source in (litellm_metadata, metadata, params_litellm_metadata, params_metadata)
+                if call_model_info and source.get("model_group") and source.get("model_info") == call_model_info
+            ),
+            None,
         )
+        model_group: Final = (
+            str(routed_source["model_group"])
+            if routed_source is not None
+            else next((str(source["model_group"]) for source in metadata_sources if source.get("model_group")), None)
+        )
+        caching_group_sources: Final = (routed_source,) if routed_source is not None else metadata_sources
         caching_group: Final = next(
-            (group for source in metadata_sources if (group := self._get_caching_group(source, model_group))), None
+            (group for source in caching_group_sources if (group := self._get_caching_group(source, model_group))), None
         )
         model_id_suffix: Final = (
             f" model_id: {kwargs['model_id']}" if "model_id" in kwargs and kwargs["model_id"] is not None else ""

litellm/completion_extras/litellm_responses_transformation/transformation.py

@@ -36,6 +36,7 @@ from litellm.completion_extras.litellm_responses_transformation.history import (
 from litellm.integrations.anthropic_cache_control_hook import supports_openai_prompt_cache_breakpoint
 from litellm.litellm_core_utils.hidden_params import get_hidden_params, get_or_create_hidden_params
 from litellm.litellm_core_utils.prompt_templates.common_utils import (
+    prompt_cache_breakpoint_for_wire,
     responses_reasoning_items_from_thinking_blocks,
     with_prompt_cache_breakpoint,
 )
@@ -61,7 +62,6 @@ from litellm.types.llms.openai import (
     ChatCompletionToolCallChunk,
     ChatCompletionToolCallFunctionChunk,
     ChatCompletionToolParamFunctionChunk,
-    PromptCacheBreakpoint,
     Reasoning,
     ResponsesAPIOptionalRequestParams,
     ResponsesAPIResponse,
@@ -631,19 +631,6 @@ def _map_incomplete_reason_to_finish_reason(incomplete_reason: str | None) -> Li
     return "length"
 
 
-_PROMPT_CACHE_BREAKPOINT: Final = TypeAdapter(PromptCacheBreakpoint)
-
-
-def _prompt_cache_breakpoint_for_wire(marker: object, drop_params: bool) -> object:
-    if marker is None or not drop_params:
-        return marker
-    try:
-        return _PROMPT_CACHE_BREAKPOINT.validate_python(marker)
-    except ValidationError:
-        verbose_logger.debug("Chat provider: dropping malformed prompt_cache_breakpoint %r under drop_params", marker)
-        return None
-
-
 def _input_file_from_file_value(file_value: object) -> dict[str, object]:
     if not isinstance(file_value, dict):
         return {"type": "input_file"}
@@ -1679,7 +1666,7 @@ class LiteLLMResponsesTransformationHandler(CompletionTransformationBridge):
                     if original_type == "text":
                         converted = with_prompt_cache_breakpoint(
                             self._convert_content_str_to_input_text(item.get("text", ""), role),
-                            _prompt_cache_breakpoint_for_wire(item.get("prompt_cache_breakpoint"), drop_params),
+                            prompt_cache_breakpoint_for_wire(item.get("prompt_cache_breakpoint"), drop_params),
                         )
                         result.append(converted)
                         verbose_logger.debug("Chat provider:   text -> %s", converted)
@@ -1692,14 +1679,14 @@ class LiteLLMResponsesTransformationHandler(CompletionTransformationBridge):
                                     cast(ChatCompletionImageObject, item), role
                                 ),
                             ),
-                            _prompt_cache_breakpoint_for_wire(item.get("prompt_cache_breakpoint"), drop_params),
+                            prompt_cache_breakpoint_for_wire(item.get("prompt_cache_breakpoint"), drop_params),
                         )
                         result.append(converted)
                         verbose_logger.debug("Chat provider:   image_url -> %s", converted)
                     elif original_type == "input_audio":
                         converted = with_prompt_cache_breakpoint(
                             {"type": "input_audio", "input_audio": item.get("input_audio")},
-                            _prompt_cache_breakpoint_for_wire(item.get("prompt_cache_breakpoint"), drop_params),
+                            prompt_cache_breakpoint_for_wire(item.get("prompt_cache_breakpoint"), drop_params),
                         )
                         result.append(converted)
                         verbose_logger.debug("Chat provider:   input_audio -> %s", converted)
@@ -1715,7 +1702,7 @@ class LiteLLMResponsesTransformationHandler(CompletionTransformationBridge):
                                 _input_file_from_file_value(
                                     cast("ChatCompletionFileObject", item).get("file"),  # cast-ok: type tag checked
                                 ),
-                                _prompt_cache_breakpoint_for_wire(item.get("prompt_cache_breakpoint"), drop_params),
+                                prompt_cache_breakpoint_for_wire(item.get("prompt_cache_breakpoint"), drop_params),
                             )
                             result.append(converted)
                             verbose_logger.debug("Chat provider:   file -> %s", converted)
@@ -1739,7 +1726,7 @@ class LiteLLMResponsesTransformationHandler(CompletionTransformationBridge):
                             # Default to input_text for unknown types
                             converted = with_prompt_cache_breakpoint(
                                 self._convert_content_str_to_input_text(str(item.get("text", item)), role),
-                                _prompt_cache_breakpoint_for_wire(item.get("prompt_cache_breakpoint"), drop_params),
+                                prompt_cache_breakpoint_for_wire(item.get("prompt_cache_breakpoint"), drop_params),
                             )
                             result.append(converted)
                             verbose_logger.debug("Chat provider:   unknown(%s) -> %s", original_type, converted)

litellm/constants.py

@@ -158,7 +158,7 @@ DEFAULT_REPLICATE_POLLING_DELAY_SECONDS: Final = get_env_int("DEFAULT_REPLICATE_
 DEFAULT_IMAGE_TOKEN_COUNT: Final = get_env_int("DEFAULT_IMAGE_TOKEN_COUNT", 250)
 HF_CONFIG_FETCH_TIMEOUT_SECONDS: Final = 10.0
 HUGGINGFACE_TOKENIZER_LOAD_RETRY_SECONDS: Final = get_env_int_in_range(
-    "HUGGINGFACE_TOKENIZER_LOAD_RETRY_SECONDS", 300, 1, 86400
+    "HUGGINGFACE_TOKENIZER_LOAD_RETRY_SECONDS", 60, 1, 86400
 )
 
 # Maximum wall-clock seconds a streaming response is allowed to run.
@@ -627,6 +627,7 @@ STS_CREDENTIAL_EXPIRY_SAFETY_MARGIN_SECONDS: Final = 60
 BEDROCK_MIN_THINKING_BUDGET_TOKENS: Final = get_env_int("BEDROCK_MIN_THINKING_BUDGET_TOKENS", 1024)
 # Anthropic's Messages API rejects thinking.budget_tokens < 1024.
 ANTHROPIC_MIN_THINKING_BUDGET_TOKENS: Final = 1024
+ANTHROPIC_BILLING_METADATA_PREFIX: Final = "x-anthropic-billing-header:"
 REPLICATE_POLLING_DELAY_SECONDS: Final = get_env_float("REPLICATE_POLLING_DELAY_SECONDS", 0.5)
 DEFAULT_ANTHROPIC_CHAT_MAX_TOKENS: Final = get_env_int("DEFAULT_ANTHROPIC_CHAT_MAX_TOKENS", 128000)
 DEFAULT_OCI_CHAT_MAX_TOKENS: Final = 4096

litellm/cost_calculator.py

@@ -22,7 +22,7 @@ from litellm.constants import (
     OPENAI_LIVE_TERMINAL_RESPONSE_EVENT_TYPES,
     REALTIME_METERED_USAGE_KEY,
 )
-from litellm.litellm_core_utils.core_helpers import as_str_mapping
+from litellm.litellm_core_utils.core_helpers import as_str_mapping, billed_usage_cost
 from litellm.litellm_core_utils.cost_discount import resolve_cost_discount
 from litellm.litellm_core_utils.decisions_pricing import decisions_pricing_fields, is_decisions_call_type
 from litellm.litellm_core_utils.hidden_params import HIDDEN_PARAMS_ATTR
@@ -50,6 +50,7 @@ from litellm.litellm_core_utils.llm_cost_calc.utils import (
     parse_prompt_tokens_details,
     select_cost_metric_for_model,
 )
+from litellm.litellm_core_utils.llm_response_utils.response_metadata import response_ms_since_logging_start
 from litellm.llms.anthropic.cost_calculation import (
     cost_per_token as anthropic_cost_per_token,
 )
@@ -1396,9 +1397,9 @@ def _apply_cost_margin(
 
 def _store_cost_breakdown_in_logging_obj(
     litellm_logging_obj: LitellmLoggingObject | None,
-    prompt_tokens_cost_usd_dollar: float,
-    completion_tokens_cost_usd_dollar: float,
-    cost_for_built_in_tools_cost_usd_dollar: float,
+    prompt_tokens_cost_usd_dollar: float | None,
+    completion_tokens_cost_usd_dollar: float | None,
+    cost_for_built_in_tools_cost_usd_dollar: float | None,
     total_cost_usd_dollar: float,
     additional_costs: dict | None = None,
     original_cost: float | None = None,
@@ -1500,9 +1501,45 @@ def _response_time_ms_for_cost(
         return float(stamped)
     if total_time:
         return total_time
-    if litellm_logging_obj is not None:
-        return litellm_logging_obj.get_response_ms()
-    return 0.0
+    if litellm_logging_obj is None:
+        return 0.0
+    reported_ms: Final = response_ms_since_logging_start(litellm_logging_obj)
+    if reported_ms is not None and litellm_logging_obj.stream is not True:
+        return reported_ms
+    return litellm_logging_obj.get_response_ms()
+
+
+def _adjusted_provider_billed_cost(
+    reported_cost: float,
+    custom_llm_provider: str | None,
+    model: str | None,
+    region_name: str | None,
+    litellm_logging_obj: LitellmLoggingObject | None,
+    service_tier: str | None,
+    data_residency: str | None,
+) -> float:
+    discounted_cost, discount_percent, discount_amount = _apply_cost_discount(
+        base_cost=reported_cost, custom_llm_provider=custom_llm_provider, model=model, region_name=region_name
+    )
+    adjusted_cost, margin_percent, margin_fixed_amount, margin_total_amount = _apply_cost_margin(
+        base_cost=discounted_cost, custom_llm_provider=custom_llm_provider
+    )
+    _store_cost_breakdown_in_logging_obj(
+        litellm_logging_obj=litellm_logging_obj,
+        prompt_tokens_cost_usd_dollar=None,
+        completion_tokens_cost_usd_dollar=None,
+        cost_for_built_in_tools_cost_usd_dollar=None,
+        total_cost_usd_dollar=adjusted_cost,
+        original_cost=reported_cost,
+        discount_percent=discount_percent,
+        discount_amount=discount_amount,
+        margin_percent=margin_percent,
+        margin_fixed_amount=margin_fixed_amount,
+        margin_total_amount=margin_total_amount,
+        service_tier=service_tier,
+        data_residency=data_residency,
+    )
+    return adjusted_cost
 
 
 def completion_cost(
@@ -1616,6 +1653,9 @@ def completion_cost(
             )
 
         explicit_pricing: Final = custom_pricing is True or base_model is not None
+        caller_set_prices: Final = (
+            explicit_pricing or custom_cost_per_token is not None or custom_cost_per_second is not None
+        )
         custom_priced_request_model: Final = model if custom_pricing is True else None
         configured_cost_per_query: Final = (
             _configured_search_cost_per_query(litellm_logging_obj) if custom_pricing else None
@@ -1666,6 +1706,7 @@ def completion_cost(
             raw_candidates if custom_pricing is True else _priced_candidates_before_unpriced_rules(raw_candidates)
         )
 
+        priced_response: Final[object] = completion_response
         for idx, (candidate_model, candidate_pricing_provider) in enumerate(potential_model_names):
             model = candidate_model
             try:
@@ -1984,6 +2025,19 @@ def completion_cost(
                     total_time=total_time,
                 )
                 model_for_discount = custom_priced_request_model or model
+                if (
+                    not caller_set_prices
+                    and (reported_cost := billed_usage_cost(priced_response, custom_llm_provider)) is not None
+                ):
+                    return _adjusted_provider_billed_cost(
+                        reported_cost=reported_cost,
+                        custom_llm_provider=custom_llm_provider,
+                        model=model_for_discount,
+                        region_name=region_name if isinstance(region_name, str) else None,
+                        litellm_logging_obj=litellm_logging_obj,
+                        service_tier=service_tier,
+                        data_residency=data_residency,
+                    )
                 # Calculate cost based on prompt_tokens, completion_tokens
                 if (
                     "togethercomputer" in model or "together_ai" in model or custom_llm_provider == "together_ai"

litellm/images/main.py

@@ -641,6 +641,8 @@ def image_generation(
                     timeout=timeout,
                     client=custom_client,
                 )
+        elif custom_llm_provider == LlmProviders.KYMA.value:
+            raise LiteLLMUnknownProvider(model=model, custom_llm_provider=custom_llm_provider)
 
         return model_response
     except Exception as e:

litellm/integrations/anthropic_cache_control_hook.py

@@ -72,6 +72,8 @@ MAX_CACHE_CONTROL_BLOCKS: Final = 4
 CACHE_BREAKPOINT_KEYS: Final = ("cache_control", "prompt_cache_breakpoint")
 OPENAI_PROMPT_CACHE_BREAKPOINT_MIN_GPT_VERSION: Final = (5, 6)
 _GPT_VERSION_PATTERN: Final = re.compile(r"^gpt-(\d+)(?:\.(\d+))?")
+_FINE_TUNED_BASE_PATTERN: Final = re.compile(r"^ft:([^:]+)")
+_O_SERIES_PATTERN: Final = re.compile(r"^o\d+(?:-|$)")
 OPENAI_PROMPT_CACHE_BREAKPOINT_BLOCK_TYPES: Final = frozenset(
     {"text", "image", "image_url", "file", "input_audio", "input_text", "input_image", "input_file"}
 )
@@ -125,6 +127,10 @@ def _served_provider_for_model(model: str) -> str | None:
 
 
 def supports_openai_prompt_cache_breakpoint(model: str, custom_llm_provider: str | None = None) -> bool:
+    return openai_prompt_cache_breakpoint_support(model, custom_llm_provider) is True
+
+
+def openai_prompt_cache_breakpoint_support(model: str, custom_llm_provider: str | None = None) -> bool | None:
     hosted_flag: Final = (
         None
         if custom_llm_provider is None
@@ -135,9 +141,19 @@ def supports_openai_prompt_cache_breakpoint(model: str, custom_llm_provider: str
     model_map_flag: Final = _model_map_prompt_cache_breakpoint_flag(model)
     if model_map_flag is not None:
         return model_map_flag
-    version_match: Final = _GPT_VERSION_PATTERN.match(model.rsplit("/", 1)[-1].lower())
-    if version_match is None:
+    fine_tuned_base: Final = _FINE_TUNED_BASE_PATTERN.match(model.rsplit("/", 1)[-1])
+    if fine_tuned_base is not None:
+        return False if _named_model_breakpoint_support(fine_tuned_base.group(1)) is False else None
+    return _named_model_breakpoint_support(model)
+
+
+def _named_model_breakpoint_support(model: str) -> bool | None:
+    bare_model: Final = model.rsplit("/", 1)[-1].lower()
+    if _O_SERIES_PATTERN.match(bare_model) is not None:
         return False
+    version_match: Final = _GPT_VERSION_PATTERN.match(bare_model)
+    if version_match is None:
+        return None
     version: Final = (int(version_match.group(1)), int(version_match.group(2) or 0))
     return version >= OPENAI_PROMPT_CACHE_BREAKPOINT_MIN_GPT_VERSION
 

litellm/integrations/datadog/datadog_cost_management.py

@@ -205,11 +205,12 @@ class DatadogCostManagementLogger(CustomBatchLogger):
         # see _RESERVED_TAG_KEYS for the rationale.
         if self.cost_tag_keys:
             allow: Final = set(self.cost_tag_keys)
+            client_allow: Final = allow - {"is_silent_experiment"}
             for rt in log.get("request_tags") or []:
                 if not isinstance(rt, str) or ":" not in rt:
                     continue
                 k, _, v = rt.partition(":")
-                if k in allow and v:
+                if k in client_allow and v:
                     self._set_custom_tag(tags, k, v)
             for k, v in metadata.items():
                 if k in allow and v is not None and not isinstance(v, (dict, list)):
@@ -218,7 +219,7 @@ class DatadogCostManagementLogger(CustomBatchLogger):
                 nested = metadata.get(nested_key)
                 if isinstance(nested, dict):
                     for k, v in nested.items():
-                        if k in allow and v is not None and not isinstance(v, (dict, list)):
+                        if k in client_allow and v is not None and not isinstance(v, (dict, list)):
                             self._set_custom_tag(tags, k, str(v))
 
         return tags

litellm/integrations/langfuse/langfuse.py

@@ -833,6 +833,12 @@ class LangFuseLogger:
                 ("vertex_location", vertex_location, bool(vertex_location)),
                 ("aws_region_name", aws_region_name, bool(aws_region_name)),
                 ("cache_hit", kwargs.get("cache_hit") or False, "cache_hit" in kwargs),
+                (
+                    "is_silent_experiment",
+                    "is_silent_experiment" in allowlisted_metadata
+                    and allowlisted_metadata["is_silent_experiment"] is True,
+                    "is_silent_experiment" in allowlisted_metadata,
+                ),
             )
             enrichments: Final[Mapping[str, object]] = {
                 key: value for key, value, include in candidate_enrichments if include

litellm/integrations/prometheus.py

@@ -1173,8 +1173,8 @@ class PrometheusLogger(CustomLogger):
         # Build label filters from valid configurations
         label_filters: Final = self._build_label_filters(parsed_configs)
 
-        # Pretty print the processed configuration
-        self._pretty_print_prometheus_config(label_filters)
+        if litellm.suppress_startup_output is not True:
+            self._pretty_print_prometheus_config(label_filters)
         return label_filters
 
     def _parse_exclude_config(self) -> tuple[frozenset[str], frozenset[str]]:
@@ -5148,6 +5148,12 @@ def _get_combined_custom_metadata_from_standard_logging_payload(
         **(requester_metadata if isinstance(requester_metadata, dict) else {}),
         **(user_api_key_auth_metadata if isinstance(user_api_key_auth_metadata, dict) else {}),
         **(spend_logs_metadata if isinstance(spend_logs_metadata, dict) else {}),
+        "is_silent_experiment": (
+            "true"
+            if "is_silent_experiment" in standard_logging_metadata
+            and standard_logging_metadata["is_silent_experiment"] is True
+            else None
+        ),
     }
 
 

litellm/integrations/websearch_interception/handler.py

@@ -37,6 +37,7 @@ from litellm.integrations.websearch_interception.transformation import (
 from litellm.litellm_core_utils.agentic_loop_settings import (
     validated_max_agentic_loops,
 )
+from litellm.litellm_core_utils.core_helpers import qualify_provider_stripped_model
 from litellm.llms.base_llm.search.transformation import SearchResponse
 from litellm.types.integrations.custom_logger import (
     CHAT_COMPLETION_AGENTIC_SURFACE,
@@ -114,6 +115,18 @@ class _WebSearchSettingsView(TypedDict):
     websearch_interception_params: WebSearchInterceptionConfig
 
 
+def _caller_model(model: str, logging_obj: "LiteLLMLoggingObj | None", kwargs: Mapping[str, object]) -> str:
+    agentic_view: Final[_AgenticLoopParamsView] = {
+        "agentic_loop_params": logging_obj.model_call_details.get("agentic_loop_params", {}) if logging_obj else {}
+    }
+    recorded_model: Final = agentic_view["agentic_loop_params"].get("model")
+    custom_llm_provider: Final = kwargs.get("custom_llm_provider")
+    return qualify_provider_stripped_model(
+        recorded_model if isinstance(recorded_model, str) and recorded_model else model,
+        custom_llm_provider if isinstance(custom_llm_provider, str) else "",
+    )
+
+
 class _SearchToolLitellmParams(TypedDict, total=False):
     search_provider: ReadOnly[str | None]
 
@@ -1155,6 +1168,7 @@ class WebSearchInterceptionLogger(CustomLogger):
             tool_calls=tool_calls,
             optional_params=optional_params,
             kwargs=kwargs,
+            logging_obj=logging_obj,
             response_format=response_format,
         )
         return AgenticLoopPlan(
@@ -1181,6 +1195,7 @@ class WebSearchInterceptionLogger(CustomLogger):
             tool_calls=tool_calls,
             optional_params=optional_params,
             kwargs=kwargs,
+            logging_obj=logging_obj,
         )
         return AgenticLoopPlan(
             run_agentic_loop=True,
@@ -1195,6 +1210,7 @@ class WebSearchInterceptionLogger(CustomLogger):
         tool_calls: list[dict],
         optional_params: dict,
         kwargs: dict,
+        logging_obj: "LiteLLMLoggingObj | None",
     ) -> AgenticLoopRequestPatch:
         """Execute litellm.asearch() and build a Responses API rerun patch."""
         search_tasks: Final = [
@@ -1254,9 +1270,7 @@ class WebSearchInterceptionLogger(CustomLogger):
             }
         }
 
-        full_model_name = model
-        if "/" not in model and isinstance(kwargs.get("custom_llm_provider"), str):
-            full_model_name = f"{kwargs['custom_llm_provider']}/{model}"
+        full_model_name: Final = _caller_model(model, logging_obj, kwargs)
 
         verbose_logger.debug(
             "WebSearchInterception: Built responses request patch model=%s input_items=%d searches=%d",
@@ -1814,6 +1828,7 @@ class WebSearchInterceptionLogger(CustomLogger):
             tool_calls=tool_calls,
             optional_params=dict[str, object](optional_params),
             kwargs=dict[str, object](kwargs),
+            logging_obj=logging_obj,
             response_format=response_format,
         )
         if request_patch.messages is None:
@@ -1836,7 +1851,8 @@ class WebSearchInterceptionLogger(CustomLogger):
         messages: list[dict],
         tool_calls: list[dict],
         optional_params: dict,
-        kwargs: dict,
+        kwargs: dict[str, object],
+        logging_obj: "LiteLLMLoggingObj | None",
         response_format: str = "openai",
     ) -> AgenticLoopRequestPatch:
         """Execute litellm.search() and build chat-completion rerun patch."""
@@ -1922,11 +1938,7 @@ class WebSearchInterceptionLogger(CustomLogger):
             k: v for k, v in kwargs.items() if not k.startswith("_websearch_interception") and k not in internal_params
         }
 
-        full_model_name = model
-        if "custom_llm_provider" in kwargs:
-            custom_llm_provider: Final = kwargs["custom_llm_provider"]
-            if not model.startswith(custom_llm_provider) and "/" not in model:
-                full_model_name = f"{custom_llm_provider}/{model}"
+        full_model_name: Final = _caller_model(model, logging_obj, kwargs)
 
         verbose_logger.debug(
             "WebSearchInterception: Built chat completion request patch model=%s messages=%d",

litellm/interactions/litellm_responses_transformation/handler.py

@@ -103,6 +103,7 @@ class LiteLLMResponsesInteractionsHandler:
         return LiteLLMResponsesInteractionsConfig.transform_responses_response_to_interactions_response(
             responses_response=responses_api_response,
             model=model,
+            custom_llm_provider=custom_llm_provider,
         )
 
     async def async_interactions_api_handler(
@@ -138,4 +139,5 @@ class LiteLLMResponsesInteractionsHandler:
         return LiteLLMResponsesInteractionsConfig.transform_responses_response_to_interactions_response(
             responses_response=responses_api_response,
             model=model,
+            custom_llm_provider=responses_request.get("custom_llm_provider"),
         )

litellm/interactions/litellm_responses_transformation/transformation.py

@@ -12,6 +12,7 @@ from typing import Final, cast
 
 from pydantic import BaseModel
 
+from litellm.litellm_core_utils.core_helpers import billed_usage_cost
 from litellm.types.interactions import (
     InteractionInput,
     InteractionsAPIOptionalRequestParams,
@@ -190,6 +191,7 @@ class LiteLLMResponsesInteractionsConfig:
     def transform_responses_response_to_interactions_response(
         responses_response: ResponsesAPIResponse,
         model: str | None = None,
+        custom_llm_provider: str | None = None,
     ) -> InteractionsAPIResponse:
         """
         Transform a Responses API response to an Interactions API response.
@@ -266,9 +268,11 @@ class LiteLLMResponsesInteractionsConfig:
         # (total_input_tokens, total_output_tokens)
         usage: Final[object] = getattr(responses_response, "usage", None)
         if usage:
+            reported_cost: Final = billed_usage_cost(responses_response, custom_llm_provider)
             interactions_response_dict["usage"] = {
                 "total_input_tokens": getattr(usage, "input_tokens", 0),
                 "total_output_tokens": getattr(usage, "output_tokens", 0),
+                **({} if reported_cost is None else {"cost": reported_cost}),
             }
 
         # Add updated (same as created for now)

litellm/litellm_core_utils/chat_completion_agentic_loop.py

@@ -15,6 +15,7 @@ from litellm.litellm_core_utils.agentic_loop_settings import (
     DEFAULT_MAX_AGENTIC_LOOPS,
     validated_max_agentic_loops,
 )
+from litellm.litellm_core_utils.core_helpers import qualify_agentic_followup_model
 from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObject
 from litellm.llms.base_llm.base_model_iterator import MockResponseIterator
 from litellm.types.integrations.custom_logger import (
@@ -182,9 +183,7 @@ async def _execute_chat_completion_agentic_plan(
     if patch.messages is None:
         raise ValueError("Agentic loop plan missing patched messages")
 
-    full_model_name = patch.model or model
-    if "/" not in full_model_name:
-        full_model_name = f"{custom_llm_provider}/{full_model_name}"
+    full_model_name: Final = qualify_agentic_followup_model(patch.model, model, custom_llm_provider)
 
     optional_params_for_followup: Final = {**optional_params, **patch.optional_params}
     if patch.tools is not None:

litellm/litellm_core_utils/core_helpers.py

@@ -2,6 +2,7 @@
 ## Helper utilities
 import copy
 import logging
+import math
 import re
 from collections.abc import Collection, Iterable, Iterator, Mapping
 from types import MappingProxyType
@@ -34,6 +35,49 @@ def is_codex_user_agent(user_agent: str) -> bool:
     return bool(_CODEX_CLIENT_PREFIX_RE.match(user_agent))
 
 
+def qualify_provider_stripped_model(model: str, custom_llm_provider: str) -> str:
+    """Put the provider prefix back on a provider-stripped model."""
+    if not custom_llm_provider or model.startswith(f"{custom_llm_provider}/"):
+        return model
+    if _already_routes_to(model, custom_llm_provider):
+        return model
+    return f"{custom_llm_provider}/{model}"
+
+
+def _already_routes_to(model: str, custom_llm_provider: str) -> bool:
+    from litellm.litellm_core_utils.get_llm_provider_logic import inferred_provider
+
+    return _leads_with_provider(model) and inferred_provider(model) == custom_llm_provider
+
+
+def _leads_with_provider(model: str) -> bool:
+    leading_segment, separator, _ = model.partition("/")
+    return bool(separator) and names_a_provider(leading_segment)
+
+
+def names_a_provider(candidate: str) -> bool:
+    """Whether a model name's leading segment is a provider rather than part of the model."""
+    import litellm
+    from litellm.litellm_core_utils.get_llm_provider_logic import is_registered_custom_provider
+    from litellm.llms.openai_like.json_loader import JSONProviderRegistry
+
+    provider_names: Final = {getattr(provider, "value", provider) for provider in litellm.provider_list}
+    return (
+        candidate in provider_names
+        or JSONProviderRegistry.exists(candidate)
+        or is_registered_custom_provider(candidate)
+    )
+
+
+def qualify_agentic_followup_model(patch_model: str | None, model: str, custom_llm_provider: str) -> str:
+    """Resolve the model an agentic follow-up re-dispatches, keeping a hook's own provider."""
+    if not patch_model or patch_model == model:
+        return qualify_provider_stripped_model(model, custom_llm_provider)
+    if _leads_with_provider(patch_model):
+        return patch_model
+    return qualify_provider_stripped_model(patch_model, custom_llm_provider)
+
+
 def safe_divide_seconds(seconds: float, denominator: float, default: float | None = None) -> float | None:
     """
     Safely divide seconds by denominator, handling zero division.
@@ -973,6 +1017,16 @@ def chunk_usage(chunk: object) -> object:
     return getattr(chunk, "usage", None)
 
 
+USAGE_COST_BILLED_PROVIDERS: Final = frozenset({"kyma"})
+
+
+def billed_usage_cost(response: object, custom_llm_provider: str | None) -> float | None:
+    if custom_llm_provider not in USAGE_COST_BILLED_PROVIDERS:
+        return None
+    cost: Final = provider_reported_usage_cost(_usage_field(response, "usage"), custom_llm_provider)
+    return cost if cost is not None and math.isfinite(cost) and cost >= 0 else None
+
+
 _MESSAGES_REPORTED_COST_PROVIDERS: Final = frozenset({"opencode_go"})
 
 

litellm/litellm_core_utils/internal_call_metadata.py

@@ -120,7 +120,7 @@ def is_unbilled_non_inference_call(
     origin: Final = metadata.get(INTERNAL_CALL_ORIGIN_METADATA_KEY) if metadata is not None else None
     if origin == BACKGROUND_RESPONSE_COST_POLL_CALL_ORIGIN:
         return False
-    if not is_background_response(response):
+    if not is_background_response(_finished_response(response)):
         return True
     return origin == BACKGROUND_RESPONSE_STATUS_POLL_CALL_ORIGIN or _is_managed_background_call(metadata)
 

litellm/litellm_core_utils/litellm_logging.py

@@ -23,7 +23,7 @@ from httpx import Response, Timeout
 from pydantic import BaseModel, JsonValue, TypeAdapter
 
 import litellm
-from litellm._internal_context import post_response_phase
+from litellm._internal_context import in_mirrored_call, outside_mirrored_call, post_response_phase
 from litellm._logging import (
     is_debugging_on,
     redact_string,
@@ -162,9 +162,10 @@ from litellm.types.llms.openai import (
 from litellm.types.mcp import MCPPostCallResponseObject
 from litellm.types.prompts.init_prompts import PromptSpec
 from litellm.types.rerank import RerankResponse
-from litellm.types.router import RouterRateLimitError
+from litellm.types.router import RouterRateLimitError, SilentExperimentFlag
 from litellm.types.utils import (
     DEPLOYMENT_SCOPED_PRICING_FIELDS,
+    AgenticLoopParams,
     CachingDetails,
     CallTypes,
     CostBreakdown,
@@ -314,7 +315,7 @@ else:
 _in_memory_loggers: Final[list[CustomLogger]] = []
 
 _STANDARD_LOGGING_METADATA_RESOLVED_KEYS: Final[frozenset[str]] = frozenset(
-    ("used_client_oauth_token", "usage_object", KUBERNETES_POD_ROUTING_KEY)
+    ("used_client_oauth_token", "usage_object", KUBERNETES_POD_ROUTING_KEY, "is_silent_experiment")
 )
 _STANDARD_LOGGING_METADATA_KEYS: Final[frozenset[str]] = (
     frozenset(StandardLoggingMetadata.__annotations__.keys()) - _STANDARD_LOGGING_METADATA_RESOLVED_KEYS
@@ -808,6 +809,7 @@ class Logging(LiteLLMLoggingBaseClass):
             litellm_params = scrub_sensitive_keys_in_metadata(litellm_params)
 
         self.litellm_params = litellm_params
+        self.is_silent_experiment: Final = in_mirrored_call()
 
         # Initialize cost breakdown field
         self.cost_breakdown: CostBreakdown | None = None
@@ -1158,6 +1160,9 @@ class Logging(LiteLLMLoggingBaseClass):
         if "custom_llm_provider" in self.model_call_details:
             self.custom_llm_provider = self.model_call_details["custom_llm_provider"]
 
+    def record_agentic_loop_params(self, params: AgenticLoopParams) -> None:
+        self.model_call_details["agentic_loop_params"] = params
+
     def record_applied_timeout(self, timeout: float | Timeout) -> None:
         self.applied_timeout = timeout
         self.litellm_params["timeout"] = timeout
@@ -1937,10 +1942,10 @@ class Logging(LiteLLMLoggingBaseClass):
 
     def set_cost_breakdown(
         self,
-        input_cost: float,
-        output_cost: float,
+        input_cost: float | None,
+        output_cost: float | None,
         total_cost: float,
-        cost_for_built_in_tools_cost_usd_dollar: float,
+        cost_for_built_in_tools_cost_usd_dollar: float | None,
         additional_costs: dict | None = None,
         original_cost: float | None = None,
         discount_percent: float | None = None,
@@ -1979,14 +1984,17 @@ class Logging(LiteLLMLoggingBaseClass):
 
         self.billed_token_rates = billed_token_rates
         self.cost_breakdown = CostBreakdown(
-            input_cost=input_cost,
-            output_cost=output_cost,
             total_cost=total_cost,
-            tool_usage_cost=cost_for_built_in_tools_cost_usd_dollar,
             service_tier=service_tier,
             data_residency=data_residency,
             vertex_location=vertex_location,
         )
+        if input_cost is not None:
+            self.cost_breakdown["input_cost"] = input_cost
+        if output_cost is not None:
+            self.cost_breakdown["output_cost"] = output_cost
+        if cost_for_built_in_tools_cost_usd_dollar is not None:
+            self.cost_breakdown["tool_usage_cost"] = cost_for_built_in_tools_cost_usd_dollar
         if cache_read_cost is not None and cache_read_cost > 0:
             self.cost_breakdown["cache_read_cost"] = cache_read_cost
         if cache_creation_cost is not None and cache_creation_cost > 0:
@@ -3249,7 +3257,7 @@ class Logging(LiteLLMLoggingBaseClass):
         """Restores trace_id/session_id contextvars once this attempt's own success
         logging (including any nested calls its callbacks trigger) is fully done."""
         try:
-            with post_response_phase():
+            with post_response_phase(), outside_mirrored_call():
                 return self._success_handler_body(
                     result=result,
                     start_time=start_time,
@@ -3749,7 +3757,7 @@ class Logging(LiteLLMLoggingBaseClass):
         """Restores trace_id/session_id contextvars once this attempt's own success
         logging (including any nested calls its callbacks trigger) is fully done."""
         try:
-            with post_response_phase():
+            with post_response_phase(), outside_mirrored_call():
                 return await self._async_success_handler_body(
                     result=result, start_time=start_time, end_time=end_time, cache_hit=cache_hit, **kwargs
                 )
@@ -4266,13 +4274,14 @@ class Logging(LiteLLMLoggingBaseClass):
         ``for_async_request`` marks the sync run of a failure whose async run also logs it, so a
         CustomLogger that gets ``async_log_failure_event`` from that async run skips its sync hook."""
         try:
-            return self._failure_handler_body(
-                exception=exception,
-                traceback_exception=traceback_exception,
-                start_time=start_time,
-                end_time=end_time,
-                for_async_request=for_async_request,
-            )
+            with outside_mirrored_call():
+                return self._failure_handler_body(
+                    exception=exception,
+                    traceback_exception=traceback_exception,
+                    start_time=start_time,
+                    end_time=end_time,
+                    for_async_request=for_async_request,
+                )
         finally:
             self._restore_correlation_context()
 
@@ -4465,12 +4474,13 @@ class Logging(LiteLLMLoggingBaseClass):
         """Restores trace_id/session_id contextvars once this attempt's own failure
         logging (including any nested calls its callbacks trigger) is fully done."""
         try:
-            return await self._async_failure_handler_body(
-                exception=exception,
-                traceback_exception=traceback_exception,
-                start_time=start_time,
-                end_time=end_time,
-            )
+            with outside_mirrored_call():
+                return await self._async_failure_handler_body(
+                    exception=exception,
+                    traceback_exception=traceback_exception,
+                    start_time=start_time,
+                    end_time=end_time,
+                )
         finally:
             self._restore_correlation_context()
 
@@ -6559,6 +6569,7 @@ class StandardLoggingPayloadSetup:
         start_time: dt_object | None = None,
         response_id: str | None = None,
         custom_llm_provider: str | None = None,
+        is_silent_experiment: bool = False,
     ) -> StandardLoggingMetadata:
         """
         Clean and filter the metadata dictionary to include only the specified keys in StandardLoggingMetadata.
@@ -6632,6 +6643,7 @@ class StandardLoggingPayloadSetup:
                 proxy_stamped_used_client_oauth_token(metadata, litellm_params),
                 custom_llm_provider,
             ),
+            **(SilentExperimentFlag(is_silent_experiment=True) if is_silent_experiment else SilentExperimentFlag()),
         )
         if isinstance(metadata, dict):
             for key in metadata.keys() & _STANDARD_LOGGING_METADATA_KEYS:
@@ -7432,6 +7444,7 @@ def get_standard_logging_object_payload(
             start_time=start_time,
             response_id=id,
             custom_llm_provider=selected_provider if isinstance(selected_provider, str) else None,
+            is_silent_experiment=logging_obj.is_silent_experiment is True,
         )
         _request_body: Final = proxy_server_request.get("body", {})
         end_user_id: Final = clean_metadata["user_api_key_end_user_id"] or _request_body.get(
@@ -7716,9 +7729,8 @@ def get_standard_logging_metadata(
     )
     if isinstance(metadata, dict):
         # Update the clean_metadata with values from input metadata that match StandardLoggingMetadata fields
-        for key in StandardLoggingMetadata.__annotations__:
-            if key in metadata:
-                clean_metadata[key] = metadata[key]
+        for key in (metadata.keys() & StandardLoggingMetadata.__annotations__.keys()) - {"is_silent_experiment"}:
+            clean_metadata[key] = metadata[key]
 
         if metadata.get("user_api_key") is not None:
             if is_valid_sha256_hash(str(metadata.get("user_api_key"))):

litellm/litellm_core_utils/llm_response_utils/response_metadata.py

@@ -27,6 +27,15 @@ def _timing_window_start(
     return start_time, False
 
 
+def response_ms_since_logging_start(logging_obj: LiteLLMLoggingObject) -> float | None:
+    reported_ms: Final = logging_obj.response_timing_metrics.get("_response_ms")
+    if reported_ms is None:
+        return None
+    logging_start: Final = logging_obj.start_time
+    window_start: Final = _timing_window_start(logging_start, logging_obj)[0]
+    return reported_ms - max(0.0, logging_start.timestamp() - window_start.timestamp()) * 1000
+
+
 def _union_duration_ms(windows: object, lower: float, upper: float) -> float | None:
     if not isinstance(windows, (list, tuple)):
         return None

litellm/litellm_core_utils/prompt_templates/common_utils.py

@@ -17,7 +17,7 @@ from pathlib import Path
 from types import MappingProxyType
 from typing import TYPE_CHECKING, Any, Final, Literal, TypeAlias, TypeVar, cast
 
-from pydantic import TypeAdapter
+from pydantic import TypeAdapter, ValidationError
 
 import litellm
 from litellm import verbose_logger
@@ -2073,6 +2073,19 @@ def with_prompt_cache_breakpoint(target: _MarkedT, marker: object) -> _MarkedT:
     return cast(_MarkedT, marked)  # cast-ok: same block shape as the input plus the marker key
 
 
+_PROMPT_CACHE_BREAKPOINT: Final = TypeAdapter(PromptCacheBreakpoint)
+
+
+def prompt_cache_breakpoint_for_wire(marker: object, drop_params: bool) -> object:
+    if marker is None or not drop_params:
+        return marker
+    try:
+        return _PROMPT_CACHE_BREAKPOINT.validate_python(marker)
+    except ValidationError:
+        verbose_logger.debug("Dropping malformed prompt_cache_breakpoint %r under drop_params", marker)
+        return None
+
+
 def cache_control_as_prompt_cache_breakpoint(block: _MarkedT) -> _MarkedT:
     """Anthropic's ``cache_control`` mark rewritten as OpenAI's ``prompt_cache_breakpoint`` on the same block."""
     if block.get("cache_control") is None:

litellm/litellm_core_utils/streaming_chunk_builder_utils.py

@@ -12,7 +12,11 @@ from typing_extensions import ReadOnly, Required
 
 import litellm
 from litellm._logging import verbose_logger
-from litellm.litellm_core_utils.core_helpers import chunk_usage, provider_reported_usage_cost
+from litellm.litellm_core_utils.core_helpers import (
+    USAGE_COST_BILLED_PROVIDERS,
+    chunk_usage,
+    provider_reported_usage_cost,
+)
 from litellm.litellm_core_utils.hidden_params import HIDDEN_PARAMS_ATTR, set_hidden_params
 from litellm.litellm_core_utils.llm_cost_calc.guardrail_cost import BEDROCK_TRACE_FIELDS, object_items, object_mapping
 from litellm.litellm_core_utils.prompt_templates.common_utils import (
@@ -1352,7 +1356,7 @@ def stream_chunk_builder_text_completion(
     return TextCompletionResponse(**response)
 
 
-_CALCULATOR_PRICED_REPORTED_COST_PROVIDERS: Final = frozenset({LlmProviders.XAI.value})
+_CALCULATOR_PRICED_REPORTED_COST_PROVIDERS: Final = frozenset({LlmProviders.XAI.value}) | USAGE_COST_BILLED_PROVIDERS
 
 
 def _logged_provider(logging_obj: Optional["Logging"]) -> object:

litellm/litellm_core_utils/token_counter.py

@@ -611,7 +611,7 @@ T_ParamSpec: Final = ParamSpec("T_ParamSpec")
 T_Retval = TypeVar("T_Retval")
 _COUNT_OFFLOAD_LIMITER: Final = anyio.lowlevel.RunVar[anyio.CapacityLimiter]("litellm_count_offload_limiter")
 _ANTHROPIC_TOOL_KEYS_MAPPED_BY_BRIDGE: Final = frozenset(
-    ("name", "input_schema", "description", "cache_control", "strict", "type", "eager_input_streaming")
+    ("name", "input_schema", "description", "cache_control", "strict", "type", "eager_input_streaming", "custom")
 )
 
 

litellm/llms/anthropic/chat/guardrail_translation/handler.py

@@ -432,27 +432,30 @@ class AnthropicMessagesHandler(BaseTranslation):
         message, _ = serialize_http_exception_detail(exc.detail)
         return tuple(anthropic_sse_error_frames(message))
 
-    def _standalone_block_chunks(self, exc: "ModifyResponseException") -> list[bytes]:
+    def build_block_response(self, exc: "ModifyResponseException") -> "AnthropicMessagesResponse":
         import uuid
 
-        from litellm.llms.anthropic.messages.native.fake_stream_iterator import (
-            FakeAnthropicMessagesStreamIterator,
-        )
         from litellm.llms.base_llm.guardrail_translation.utils import (
             blocked_response_usage,
         )
         from litellm.types.utils import AnthropicMessagesResponse
 
-        block_response: Final = AnthropicMessagesResponse(
+        return AnthropicMessagesResponse(
             id=f"msg_{uuid.uuid4()}",
             type="message",
             role="assistant",
             content=[{"type": "text", "text": exc.message}],
             model=exc.model,
             stop_reason="end_turn",
-            usage=blocked_response_usage(getattr(exc, "original_response", None)),
+            usage=blocked_response_usage(exc.original_response),
         )
-        return list(FakeAnthropicMessagesStreamIterator(response=block_response))
+
+    def _standalone_block_chunks(self, exc: "ModifyResponseException") -> list[bytes]:
+        from litellm.llms.anthropic.messages.native.fake_stream_iterator import (
+            FakeAnthropicMessagesStreamIterator,
+        )
+
+        return list(FakeAnthropicMessagesStreamIterator(response=self.build_block_response(exc)))
 
     def _block_continuation_chunks(
         self, exc: "ModifyResponseException", responses_so_far: Sequence[object]

litellm/llms/anthropic/chat/handler.py

@@ -17,6 +17,7 @@ import litellm
 import litellm.litellm_core_utils
 import litellm.types
 import litellm.types.utils
+from litellm.anthropic_beta_gated_fields import strip_beta_gated_request_body
 from litellm.anthropic_beta_headers_manager import (
     update_request_with_filtered_beta,
 )
@@ -506,11 +507,20 @@ class AnthropicChatCompletion(BaseLLM):
             The pair stays mutable because the streaming path rewrites it in
             place (`data["stream"] = True`) before sending.
             """
-            request_headers, data = update_request_with_filtered_beta(
+            request_headers, filtered_data = update_request_with_filtered_beta(
                 headers=headers,
                 request_data=request_data,
                 provider=custom_llm_provider,
             )
+            data: Final = dict(
+                strip_beta_gated_request_body(
+                    filtered_data,
+                    request_headers,
+                    drop_params=litellm_params.get("drop_params"),  # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType]  # litellm_params is the handler's bare dict
+                    model=model,
+                    provider=custom_llm_provider,
+                )
+            )
             logging_obj.pre_call(
                 input=messages,
                 api_key=api_key,

litellm/llms/anthropic/chat/transformation.py

@@ -12,6 +12,7 @@ from typing_extensions import ReadOnly, TypedDict
 
 import litellm
 from litellm.constants import (
+    ANTHROPIC_BILLING_METADATA_PREFIX,
     ANTHROPIC_MIN_THINKING_BUDGET_TOKENS,
     ANTHROPIC_WEB_SEARCH_TOOL_MAX_USES,
     DEFAULT_ANTHROPIC_CHAT_MAX_TOKENS,
@@ -2066,7 +2067,7 @@ class AnthropicConfig(AnthropicModelInfo, BaseConfig):
                     if not system_message_block["content"]:
                         continue
                     if self.should_strip_billing_metadata() and system_message_block["content"].startswith(
-                        "x-anthropic-billing-header:"
+                        ANTHROPIC_BILLING_METADATA_PREFIX
                     ):
                         continue
                     anthropic_system_message_content = AnthropicSystemMessageContent(
@@ -2086,7 +2087,7 @@ class AnthropicConfig(AnthropicModelInfo, BaseConfig):
                             self.should_strip_billing_metadata()
                             and _content.get("type") == "text"
                             and text_value
-                            and text_value.startswith("x-anthropic-billing-header:")
+                            and text_value.startswith(ANTHROPIC_BILLING_METADATA_PREFIX)
                         ):
                             continue
                         anthropic_system_message_content = AnthropicSystemMessageContent(

litellm/llms/anthropic/common_utils.py

@@ -51,6 +51,7 @@ from litellm.types.llms.anthropic import (
     ANTHROPIC_OAUTH_BETA_HEADER,
     ANTHROPIC_OAUTH_TOKEN_PREFIX,
     ANTHROPIC_SERVER_TOOL_USE_ID_PREFIX,
+    ANTHROPIC_THINKING_BINDING_CONTROLS_BETA_HEADER,
     ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER,
     ANTHROPIC_TOOL_SEARCH_TOOL_TYPES,
     AllAnthropicToolsValues,
@@ -121,6 +122,54 @@ _DECODED_JSON: Final = TypeAdapter(object)
 _CLAUDE_CODE_USER_AGENT_PREFIXES: Final = ("claude-cli/", "claude-code/")
 
 
+def is_claude_tool_target(custom_llm_provider: str | None, model: str) -> bool:
+    if custom_llm_provider == "anthropic":
+        return True
+    model_lower: Final = model.lower()
+    if custom_llm_provider == "bedrock":
+        return "claude" in model_lower or ("arn:" in model_lower and ":bedrock:" in model_lower)
+    if custom_llm_provider == "azure_ai" or custom_llm_provider == "vertex_ai":
+        return "claude" in model_lower
+    return False
+
+
+def is_claude_model_target(custom_llm_provider: str | None, model: str) -> bool:
+    return "claude" in model.lower() and is_claude_tool_target(custom_llm_provider, model)
+
+
+def _hoistable_defer_loading(tool_dict: Mapping[object, object], custom: object) -> bool | None:
+    if not isinstance(custom, dict) or "defer_loading" in tool_dict:
+        return None
+    defer_loading: Final = OBJECT_MAPPING_ADAPTER.validate_python(custom).get("defer_loading")
+    return defer_loading if isinstance(defer_loading, bool) else None
+
+
+def _tool_without_custom_field(tool: object, hoist_defer_loading: bool, drop_custom: bool) -> object:
+    if not isinstance(tool, dict):
+        return tool
+
+    if "custom" not in tool:
+        return tool
+
+    tool_dict: Final = OBJECT_MAPPING_ADAPTER.validate_python(tool)
+
+    custom: Final = tool_dict["custom"]
+    without_custom: Final = {key: value for key, value in tool_dict.items() if key != "custom"}
+    defer_loading: Final = _hoistable_defer_loading(tool_dict, custom) if hoist_defer_loading else None
+    if defer_loading is None:
+        return without_custom if drop_custom else tool_dict
+
+    unhoisted: Final = {
+        key: value for key, value in OBJECT_MAPPING_ADAPTER.validate_python(custom).items() if key != "defer_loading"
+    }
+    kept_custom: Final = {"custom": unhoisted} if unhoisted and not drop_custom else {}
+    return {**without_custom, "defer_loading": defer_loading, **kept_custom}
+
+
+def tools_without_custom_field(tools: Sequence[object], hoist_defer_loading: bool, drop_custom: bool) -> list[object]:
+    return [_tool_without_custom_field(tool, hoist_defer_loading, drop_custom) for tool in tools]
+
+
 def is_anthropic_messages_url(url: str) -> bool:
     """Check whether a URL addresses Anthropic's Messages API."""
     parsed_url: Final = urlparse(url)
@@ -552,6 +601,20 @@ class AnthropicModelInfo(BaseLLMModelInfo):
             return False
         return thinking.get("type") in ("adaptive", "enabled") and thinking.get("display") == "updates"
 
+    def thinking_betas(self, thinking: AnthropicThinkingParam | None) -> tuple[str, ...]:
+        return (
+            *(
+                (ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER,)
+                if self.is_thinking_display_updates_used(thinking)
+                else ()
+            ),
+            *(
+                (ANTHROPIC_THINKING_BINDING_CONTROLS_BETA_HEADER,)
+                if isinstance(thinking, dict) and "block_binding" in thinking
+                else ()
+            ),
+        )
+
     def is_mid_conversation_tool_change_used(self, messages: Sequence[object]) -> bool:
         for message in messages:
             if message_field(message, "role") != "system":
@@ -1132,7 +1195,7 @@ class AnthropicModelInfo(BaseLLMModelInfo):
         *,
         custom_llm_provider: str,
         is_mid_conversation_output_config_used: bool = False,
-        is_thinking_display_updates_used: bool = False,
+        thinking: AnthropicThinkingParam | None = None,
         is_mid_conversation_tool_change_used: bool = False,
     ) -> list[str]:
         """
@@ -1169,13 +1232,10 @@ class AnthropicModelInfo(BaseLLMModelInfo):
         if is_mid_conversation_output_config_used:
             betas.append(ANTHROPIC_MID_CONVERSATION_OUTPUT_CONFIG_BETA_HEADER)
 
-        thinking_display_betas: Final = (
-            (ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER,) if is_thinking_display_updates_used else ()
-        )
         tool_change_betas: Final = (
             (ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER,) if is_mid_conversation_tool_change_used else ()
         )
-        return list(set(betas).union(thinking_display_betas, tool_change_betas))
+        return list(set(betas).union(self.thinking_betas(thinking), tool_change_betas))
 
     @staticmethod
     def _make_api_key_auth_header(
@@ -1237,7 +1297,7 @@ class AnthropicModelInfo(BaseLLMModelInfo):
         api_base: str | None = None,
         use_bearer_for_custom_base: bool = False,
         is_mid_conversation_output_config_used: bool = False,
-        is_thinking_display_updates_used: bool = False,
+        thinking: AnthropicThinkingParam | None = None,
         is_mid_conversation_tool_change_used: bool = False,
         wif_minted: bool = False,
     ) -> dict:
@@ -1298,7 +1358,7 @@ class AnthropicModelInfo(BaseLLMModelInfo):
             betas.update(user_anthropic_beta_headers)
 
         all_betas: Final = betas.union(
-            (ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER,) if is_thinking_display_updates_used else (),
+            self.thinking_betas(thinking),
             (ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER,) if is_mid_conversation_tool_change_used else (),
         )
 
@@ -1385,7 +1445,7 @@ class AnthropicModelInfo(BaseLLMModelInfo):
             auth_token=auth_token,
             file_id_used=file_id_used,
             is_mid_conversation_output_config_used=is_mid_conversation_output_config_used,
-            is_thinking_display_updates_used=self.is_thinking_display_updates_used(optional_params.get("thinking")),
+            thinking=optional_params.get("thinking"),
             is_mid_conversation_tool_change_used=self.is_mid_conversation_tool_change_used(messages),
             web_search_tool_used=web_search_tool_used,
             is_vertex_request=optional_params.get("is_vertex_request", False),

litellm/llms/anthropic/count_tokens/handler.py

@@ -39,6 +39,7 @@ class AnthropicCountTokensHandler(AnthropicCountTokensConfig):
         tools: list[dict[str, JsonValue]] | None = None,
         system: JsonValue = None,
         optional_params: Mapping[str, JsonValue] | None = None,
+        drop_params: bool = False,
     ) -> dict[str, JsonValue]:
         """
         Handle a CountTokens request using httpx.
@@ -69,6 +70,7 @@ class AnthropicCountTokensHandler(AnthropicCountTokensConfig):
                 tools=tools,
                 system=system,
                 optional_params=optional_params,
+                drop_params=drop_params,
             )
 
             verbose_logger.debug("Transformed request: %s", request_body)

litellm/llms/anthropic/count_tokens/token_counter.py

@@ -11,6 +11,7 @@ from pydantic import JsonValue
 import litellm
 from litellm._logging import verbose_logger
 from litellm.exceptions import AuthenticationError
+from litellm.litellm_core_utils.core_helpers import drops_unsupported_params
 from litellm.llms.anthropic.count_tokens.handler import AnthropicCountTokensHandler
 from litellm.llms.anthropic.count_tokens.transformation import (
     AnthropicCountFields,
@@ -131,6 +132,7 @@ class AnthropicTokenCounter(BaseTokenCounter):
                 tools=tools,
                 system=system,
                 optional_params=count_options,
+                drop_params=drops_unsupported_params(litellm_params.get("drop_params")),
             )
 
             if result is not None:

litellm/llms/anthropic/count_tokens/transformation.py

@@ -26,7 +26,11 @@ from litellm.litellm_core_utils.prompt_templates.mid_conversation_system import
     split_leading_system_run,
 )
 from litellm.litellm_core_utils.url_utils import join_api_base
-from litellm.llms.anthropic.common_utils import merge_anthropic_beta_headers
+from litellm.llms.anthropic.common_utils import (
+    is_claude_model_target,
+    merge_anthropic_beta_headers,
+    tools_without_custom_field,
+)
 from litellm.llms.anthropic.wif import resolve_anthropic_base
 from litellm.types.llms.openai import AllMessageValues, ChatCompletionImageObject, ValidUserMessageContentTypes
 from litellm.types.utils import TokenCountMessagesFormat, TokenCountResponse
@@ -372,6 +376,7 @@ class AnthropicCountTokensConfig:
         tools: list[dict[str, JsonValue]] | None = None,
         system: JsonValue = None,
         optional_params: Mapping[str, JsonValue] | None = None,
+        drop_params: bool = False,
     ) -> dict[str, JsonValue]:  # mutable-ok: provider transport requires JSON dictionaries
         """
         Transform request to Anthropic CountTokens format.
@@ -391,7 +396,19 @@ class AnthropicCountTokensConfig:
                     **MappingProxyType(
                         {
                             key: value
-                            for key, value in (("system", counted_system), ("tools", tools))
+                            for key, value in (
+                                ("system", counted_system),
+                                (
+                                    "tools",
+                                    None
+                                    if tools is None
+                                    else tools_without_custom_field(
+                                        tools,
+                                        hoist_defer_loading=is_claude_model_target("anthropic", model),
+                                        drop_custom=drop_params,
+                                    ),
+                                ),
+                            )
                             if value is not None
                         }
                     ),

litellm/llms/anthropic/messages/adapters/transformation.py

@@ -203,7 +203,7 @@ from litellm.types.llms.openai import (
     ChatCompletionUserMessage,
     ToolMessageContentPart,
 )
-from litellm.types.utils import Choices, ModelResponse, StreamingChoices, Usage
+from litellm.types.utils import Choices, LlmProviders, ModelResponse, StreamingChoices, Usage
 from litellm.utils import supports_mid_conversation_system
 
 from .streaming_iterator import AnthropicSSEStream, AnthropicStreamWrapper
@@ -362,6 +362,9 @@ class AnthropicAdapter:
 _BlockT: Final = TypeVar("_BlockT", bound=Mapping[str, object])
 
 
+_CLAUDE_OVER_REASONING_EFFORT_PROVIDERS: Final = frozenset({LlmProviders.KYMA.value})
+
+
 class LiteLLMAnthropicMessagesAdapter:
     def __init__(self):
         pass
@@ -817,6 +820,7 @@ class LiteLLMAnthropicMessagesAdapter:
             "name",
             "input_schema",
             "description",
+            "custom",
             "cache_control",
             "strict",
             "type",
@@ -1127,7 +1131,8 @@ class LiteLLMAnthropicMessagesAdapter:
         """Translate Anthropic thinking to either thinking or reasoning_effort.
 
         A Claude-family target keeps ``thinking`` verbatim, since every bridged provider serving one
-        speaks that param. Carrying its adaptive effort tier alongside takes two different params,
+        speaks that param, except Kyma, which serves Claude over the OpenAI dialect and takes only
+        ``reasoning_effort``, so its Claude models get the non-Claude mapping. Carrying its adaptive effort tier alongside takes two different params,
         because the two are not interchangeable at the provider mapping below.
 
         Bedrock takes ``output_config`` directly, which attaches the tier and leaves ``thinking``
@@ -1158,7 +1163,9 @@ class LiteLLMAnthropicMessagesAdapter:
         is_bedrock_target: Final = model.startswith(("bedrock/", "converse/", "invoke/")) or self.is_bedrock_arn_model(
             model
         )
-        is_claude_target: Final = self.is_anthropic_claude_model(model) or self.is_bedrock_arn_model(model)
+        is_claude_target: Final = (
+            self.is_anthropic_claude_model(model) or self.is_bedrock_arn_model(model)
+        ) and custom_llm_provider not in _CLAUDE_OVER_REASONING_EFFORT_PROVIDERS
         output_config: Final = anthropic_message_request.get("output_config")
 
         if is_claude_target:

litellm/llms/anthropic/messages/native/transformation.py

@@ -2,10 +2,22 @@ from collections.abc import AsyncIterator, Mapping, Sequence
 from typing import ClassVar, Final
 
 import httpx
+from pydantic import TypeAdapter
 
 import litellm
 from litellm._logging import verbose_logger
-from litellm.anthropic_beta_headers_manager import get_unsupported_headers, is_beta_header_supported_by_bundled_config
+from litellm.anthropic_beta_gated_fields import (
+    log_beta_gated_strips,
+    strip_beta_gated_native_request,
+    strip_beta_gated_system_messages,
+    unstripped_native_request,
+)
+from litellm.anthropic_beta_headers_manager import (
+    get_unsupported_headers,
+    has_beta_headers_config,
+    is_beta_header_supported_by_bundled_config,
+)
+from litellm.constants import ANTHROPIC_BILLING_METADATA_PREFIX
 from litellm.exceptions import AuthenticationError
 from litellm.litellm_core_utils.core_helpers import drops_unsupported_params
 from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
@@ -19,7 +31,6 @@ from litellm.types.llms.anthropic import (
     ANTHROPIC_ADVISOR_TOOL_TYPE,
     ANTHROPIC_BETA_HEADER_VALUES,
     ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER,
-    ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER,
     AnthropicMessagesRequest,
 )
 from litellm.types.llms.anthropic_messages.anthropic_response import (
@@ -29,13 +40,16 @@ from litellm.types.llms.anthropic_tool_search import get_tool_search_beta_header
 from litellm.types.router import GenericLiteLLMParams
 
 from ...common_utils import (
+    OBJECT_LIST_ADAPTER,
     AnthropicError,
     AnthropicModelInfo,
+    is_claude_model_target,
     merge_anthropic_beta_headers,
     optionally_handle_anthropic_oauth,
     requires_native_compaction_beta,
     strip_advisor_blocks_from_messages,
     strip_encrypted_reasoning_blocks_from_anthropic_messages,
+    tools_without_custom_field,
 )
 from .mid_conversation_system import (
     as_system_content_blocks,
@@ -47,6 +61,8 @@ DEFAULT_ANTHROPIC_API_VERSION: Final = "2023-06-01"
 
 _CALLER_CREDENTIAL_HEADERS: Final = frozenset({"x-api-key", "authorization"})
 
+_MESSAGE_MAPPINGS: Final = TypeAdapter(tuple[Mapping[str, object], ...])
+
 
 def _carries_caller_credential(headers: Mapping[str, str]) -> bool:
     """Whether the caller sent their own Anthropic credential, in which case this passthrough
@@ -81,6 +97,9 @@ def _ensure_stop_sequence(response: AnthropicMessagesResponse) -> AnthropicMessa
     return response
 
 
+_MESSAGES_ADAPTER: Final = TypeAdapter(list[dict[str, object]])
+
+
 class AnthropicMessagesConfig(BaseAnthropicMessagesConfig):
     _workload_identity_eligible: ClassVar[bool] = True
 
@@ -181,6 +200,16 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig):
         """
         return False
 
+    def hoists_wrapped_defer_loading(self, model: str) -> bool:
+        return is_claude_model_target(self._resolved_provider, model)
+
+    def tools_for_upstream(self, tools: Sequence[object], model: str, drop_params: object) -> Sequence[object]:
+        return tools_without_custom_field(
+            tools,
+            hoist_defer_loading=self.hoists_wrapped_defer_loading(model),
+            drop_custom=drops_unsupported_params(drop_params),
+        )
+
     @staticmethod
     def _filter_billing_headers_from_system(system_param):
         """
@@ -194,7 +223,7 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig):
         """
         if isinstance(system_param, str):
             # If it's a string and starts with billing header, filter it out
-            if system_param.startswith("x-anthropic-billing-header:"):
+            if system_param.startswith(ANTHROPIC_BILLING_METADATA_PREFIX):
                 return None
             return system_param
         elif isinstance(system_param, list):
@@ -205,7 +234,7 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig):
                     text = content_block.get("text", "")
                     content_type = content_block.get("type", "")
                     # Skip text blocks that start with billing header
-                    if content_type == "text" and text.startswith("x-anthropic-billing-header:"):
+                    if content_type == "text" and text.startswith(ANTHROPIC_BILLING_METADATA_PREFIX):
                         continue
                     filtered_list.append(content_block)
                 else:
@@ -215,7 +244,18 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig):
         else:
             return system_param
 
-    def _normalize_system_role_messages(self, anthropic_messages_request: dict, model: str) -> None:
+    def _strip_beta_gated_system_turns(
+        self, messages: Sequence[Mapping[str, object]], model: str, drop_params: object
+    ) -> Sequence[Mapping[str, object]]:
+        if not has_beta_headers_config(self._resolved_provider):
+            return messages
+        gated: Final = strip_beta_gated_system_messages(messages, frozenset(), drop_params=drop_params)
+        log_beta_gated_strips(gated.stripped, model=model, provider=self._resolved_provider)
+        return _MESSAGE_MAPPINGS.validate_python(gated.messages) if gated.stripped else messages
+
+    def _normalize_system_role_messages(
+        self, anthropic_messages_request: dict, model: str, drop_params: object
+    ) -> None:
         """Normalize ``role: "system"`` entries in ``messages`` per the Anthropic
         ``/v1/messages`` contract, which the first-party API, Bedrock Invoke,
         Vertex, and Azure Foundry all enforce identically.
@@ -263,7 +303,15 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig):
                 custom_llm_provider=self.custom_llm_provider,
                 key="supports_mid_conversation_system",
             )
-            else list(convert_mid_conversation_system_turns(messages[leading_count:]))
+            else list(
+                convert_mid_conversation_system_turns(
+                    self._strip_beta_gated_system_turns(
+                        messages[leading_count:],  # pyright: ignore[reportUnknownArgumentType]  # messages comes from the bare request dict
+                        model=model,
+                        drop_params=drop_params,
+                    )
+                )
+            )
         )
         if hoisted or remaining != messages:
             anthropic_messages_request["messages"] = remaining
@@ -322,7 +370,7 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig):
                     model=model,
                 ),
             )
-        return self._finalize_messages_headers(headers, optional_params, messages), api_base
+        return self.finalize_messages_headers(headers, optional_params, messages), api_base
 
     async def avalidate_anthropic_messages_environment(
         self,
@@ -362,7 +410,7 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig):
                     model=model,
                 ),
             )
-        return self._finalize_messages_headers(oauth_headers, optional_params, messages), api_base
+        return self.finalize_messages_headers(oauth_headers, optional_params, messages), api_base
 
     def _require_auth_header(self, auth_header: Mapping[str, str] | None, model: str) -> Mapping[str, str]:
         if auth_header is None:
@@ -396,7 +444,7 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig):
 
         return config_allows_workload_identity(self)
 
-    def _finalize_messages_headers(
+    def finalize_messages_headers(
         self,
         headers: dict,  # mutable-ok: out-param
         optional_params: dict,  # mutable-ok: out-param
@@ -605,6 +653,12 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig):
         if thinking_enabled or effort_enabled:
             optional_params.pop("temperature", None)
 
+    def strips_beta_gated_fields_from_final_headers(self) -> bool:
+        """False on legs whose final ``anthropic_beta`` list is only known after this transform (Bedrock Invoke),
+        which strip beta-gated fields themselves once that list exists, and on providers with no row in
+        ``anthropic_beta_headers_config.json``, whose filter drops every beta and so cannot say which fields fail."""
+        return has_beta_headers_config(self._resolved_provider)
+
     def transform_anthropic_messages_request(
         self,
         model: str,
@@ -700,20 +754,47 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig):
 
         # Auto-strip advisor blocks from history if advisor tool is absent.
         # Prevents Anthropic 400: advisor_tool_result in history requires advisor tool.
-        _tools: Final = anthropic_messages_optional_request_params.get("tools") or []
+        tools: Final = anthropic_messages_optional_request_params.get("tools")
+        _tools: Final = tools or []
         _has_advisor: Final = any(isinstance(t, dict) and t.get("type") == ANTHROPIC_ADVISOR_TOOL_TYPE for t in _tools)
         if not _has_advisor:
             messages = strip_advisor_blocks_from_messages(messages)
 
+        gated: Final = (
+            strip_beta_gated_native_request(
+                headers,  # pyright: ignore[reportUnknownArgumentType]  # the base signature types this as a bare dict
+                anthropic_messages_optional_request_params,  # pyright: ignore[reportUnknownArgumentType]  # the base signature types this as a bare dict
+                messages,  # pyright: ignore[reportUnknownArgumentType]  # the base signature types this as a bare list
+                drop_params=getattr(litellm_params, "drop_params", None),
+            )
+            if self.strips_beta_gated_fields_from_final_headers()
+            else unstripped_native_request(anthropic_messages_optional_request_params, messages)  # pyright: ignore[reportUnknownArgumentType]  # the base signature types these as bare dicts
+        )
+        log_beta_gated_strips(gated.stripped, model=model, provider=self._resolved_provider)
         anthropic_messages_request: Final[AnthropicMessagesRequest] = AnthropicMessagesRequest(
             messages=strip_encrypted_reasoning_blocks_from_anthropic_messages(
-                drop_non_string_tool_result_texts(messages, getattr(litellm_params, "drop_params", None))
+                drop_non_string_tool_result_texts(
+                    _MESSAGES_ADAPTER.validate_python(gated.messages) if gated.messages_stripped else messages,
+                    getattr(litellm_params, "drop_params", None),
+                )
             ),
             max_tokens=max_tokens,
             model=model,
             **anthropic_messages_optional_request_params,
         )
-        return dict(anthropic_messages_request)
+        request: Final = (
+            {**dict(anthropic_messages_request), "thinking": gated.thinking}
+            if gated.thinking_stripped
+            else dict(anthropic_messages_request)  # pyright: ignore[reportUnknownVariableType]  # built from the base signature's bare dict
+        )
+        if not isinstance(tools, list):
+            return request
+        return {
+            **request,
+            "tools": self.tools_for_upstream(
+                OBJECT_LIST_ADAPTER.validate_python(tools), model, litellm_params.drop_params
+            ),
+        }
 
     def transform_anthropic_messages_response(
         self,
@@ -844,17 +925,13 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig):
         if AnthropicModelInfo().is_tool_search_used(tools):
             beta_values.add(get_tool_search_beta_header(custom_llm_provider))
 
-        thinking_display_betas: Final = (
-            (ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER,)
-            if AnthropicModelInfo().is_thinking_display_updates_used(optional_params.get("thinking"))
-            else ()
-        )
+        thinking_betas: Final = AnthropicModelInfo().thinking_betas(optional_params.get("thinking"))
         tool_change_betas: Final = (
             (ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER,)
             if AnthropicModelInfo().is_mid_conversation_tool_change_used(messages)
             else ()
         )
-        all_beta_values: Final = beta_values.union(thinking_display_betas, tool_change_betas)
+        all_beta_values: Final = beta_values.union(thinking_betas, tool_change_betas)
         dangerous_tool_use: Final = ANTHROPIC_BETA_HEADER_VALUES.DANGEROUS_TOOL_USE_2026_09_03.value
         forwarded_beta_values: Final = (  # pyright: ignore[reportUnknownVariableType]  # derived from the bare beta set
             all_beta_values - {dangerous_tool_use}

litellm/llms/anthropic/messages/responses_adapters/handler.py

@@ -15,7 +15,11 @@ from litellm.integrations.anthropic_cache_control_hook import (
     supports_openai_prompt_cache_breakpoint,
 )
 from litellm.litellm_core_utils.core_helpers import drops_unsupported_params
-from litellm.litellm_core_utils.prompt_templates.common_utils import cache_control_as_prompt_cache_breakpoint
+from litellm.litellm_core_utils.prompt_templates.common_utils import (
+    cache_control_as_prompt_cache_breakpoint,
+    prompt_cache_breakpoint_for_wire,
+    with_prompt_cache_breakpoint,
+)
 from litellm.responses.utils import auto_reasoning_summary
 from litellm.types.integrations.custom_logger import CONVERTED_STREAM_KEYS
 from litellm.types.llms.anthropic import (
@@ -90,14 +94,28 @@ def _without_prompt_cache_breakpoint(block: Mapping[str, object]) -> Mapping[str
     return {key: value for key, value in block.items() if key != "prompt_cache_breakpoint"}
 
 
+def _cache_control_as_wire_safe_prompt_cache_breakpoint(block: Mapping[str, object]) -> Mapping[str, object]:
+    if "prompt_cache_breakpoint" not in block:
+        return cache_control_as_prompt_cache_breakpoint(block)
+    wire_marker: Final = prompt_cache_breakpoint_for_wire(block["prompt_cache_breakpoint"], drop_params=True)
+    return cache_control_as_prompt_cache_breakpoint(
+        with_prompt_cache_breakpoint(_without_prompt_cache_breakpoint(block), wire_marker)
+    )
+
+
 def _prompt_cache_block_rewrite(model: str, forwarded_kwargs: Mapping[str, object]) -> BlockRewrite | None:
     """Client cache marks become breakpoints for a deployment that reads them, and a model that takes no breakpoints
     drops raw ones under drop_params."""
+    drops_params: Final = drops_unsupported_params(forwarded_kwargs.get("drop_params"))
     if _reads_prompt_cache_breakpoint(model, forwarded_kwargs):
-        return cache_control_as_prompt_cache_breakpoint
+        return (
+            _cache_control_as_wire_safe_prompt_cache_breakpoint
+            if drops_params
+            else cache_control_as_prompt_cache_breakpoint
+        )
     if _model_takes_prompt_cache_breakpoint(model, forwarded_kwargs):
         return None
-    if drops_unsupported_params(forwarded_kwargs.get("drop_params")):
+    if drops_params:
         return _without_prompt_cache_breakpoint
     return None
 

litellm/llms/anthropic/prompt_cache_prediction.py

@@ -661,12 +661,13 @@ def prepare_native_baseline_body(request: Mapping[str, object], model: str) -> M
             TypeAdapter(tuple[str, ...]).validate_python(owned.get("additional_drop_params") or ()),
             dict(options),
         )
-        body: Final = AnthropicMessagesConfig().transform_anthropic_messages_request(
+        config: Final = AnthropicMessagesConfig()
+        body: Final = config.transform_anthropic_messages_request(
             model=resolved_model,
             messages=strip_provider_specific_fields_from_anthropic_messages(messages),
             anthropic_messages_optional_request_params=filtered,
             litellm_params=GenericLiteLLMParams(),
-            headers={},
+            headers=config.finalize_messages_headers({}, dict(filtered), list(messages)),  # pyright: ignore[reportUnknownMemberType]  # the shared helper takes and returns bare dicts
         )
         return MappingProxyType(_JSON_OBJECT.validate_python(body))
     except Exception:  # noqa: BLE001  # an unsupported hypothetical request is unavailable, never an inference failure

litellm/llms/azure/chat/gpt_reasoning_transformation.py

@@ -24,6 +24,7 @@ class AzureOpenAIGPT5Config(AzureOpenAIConfig, OpenAIGPT5Config):
 
     GPT5_SERIES_ROUTE = "gpt5_series/"
     OPT_IN_REASONING_EFFORTS: ClassVar[frozenset[str]] = frozenset(("xhigh", "none"))
+    LOOKS_UP_BASE_MODEL: ClassVar[bool] = True
 
     @classmethod
     def _model_map_lookup_name(cls, model: str) -> str:
@@ -88,7 +89,7 @@ class AzureOpenAIGPT5Config(AzureOpenAIConfig, OpenAIGPT5Config):
         drop_params: bool,
         api_version: str = "",
         *,
-        declared_reasoning_efforts: Collection[str] = (),
+        declared_reasoning_efforts: Collection[str] | None = None,
     ) -> dict:
         return OpenAIGPT5Config.map_openai_params(
             self,

litellm/llms/azure/chat/gpt_transformation.py

@@ -1,4 +1,4 @@
-from collections.abc import Mapping
+from collections.abc import Mapping, Sequence
 from types import MappingProxyType
 from typing import TYPE_CHECKING, Any, Final
 
@@ -15,6 +15,7 @@ from litellm.litellm_core_utils.prompt_templates.common_utils import (
 from litellm.litellm_core_utils.prompt_templates.factory import (
     convert_to_azure_openai_messages,
 )
+from litellm.llms.base_llm.base_utils import map_developer_role_to_system_role
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
 from litellm.llms.openai.chat.gpt_reasoning_transformation import GPT_REASONING_SERIES_MARKERS
 from litellm.types.llms.azure import (
@@ -98,6 +99,22 @@ class AzureOpenAIConfig(BaseConfig):
             if key != "self" and value is not None:
                 setattr(self.__class__, key, value)
 
+    def translate_developer_role_to_system_role(
+        self,
+        messages: Sequence[AllMessageValues],
+        *,
+        model: str,
+        custom_llm_provider: str | None,
+        api_base: str | None,
+        model_info: object,
+    ) -> Sequence[AllMessageValues]:
+        """
+        Azure OpenAI accepts `system` messages at any position, so a developer message
+        only changes role. Declared here so `AzureOpenAIGPT5Config` resolves to it
+        ahead of the OpenAI-compatible hoist it also inherits.
+        """
+        return map_developer_role_to_system_role(messages=messages)
+
     @classmethod
     def get_config(cls):
         return super().get_config()

litellm/llms/azure/responses/transformation.py

@@ -335,7 +335,7 @@ class AzureOpenAIResponsesAPIConfig(OpenAIResponsesAPIConfig):
         api_base: str,
         litellm_params: GenericLiteLLMParams,
         headers: dict,
-    ) -> tuple[str, dict]:
+    ) -> tuple[str, dict[str, str]]:
         """
         Transform the get response API request into a URL and data
 
@@ -343,7 +343,7 @@ class AzureOpenAIResponsesAPIConfig(OpenAIResponsesAPIConfig):
         - GET /v1/responses/{response_id}
         """
         get_url: Final = self._construct_url_for_response_id_in_path(api_base=api_base, response_id=response_id)
-        data: Final[dict] = {}
+        data: Final[dict[str, str]] = {}
         verbose_logger.debug("get response url=%s", get_url)
         return get_url, data
 

litellm/llms/azure_ai/anthropic/count_tokens/handler.py

@@ -13,6 +13,7 @@ from pydantic import JsonValue
 import litellm
 from litellm._logging import verbose_logger
 from litellm.litellm_core_utils.asyncify import asyncify
+from litellm.litellm_core_utils.core_helpers import drops_unsupported_params
 from litellm.llms.anthropic.common_utils import AnthropicError
 from litellm.llms.azure_ai.anthropic.count_tokens.transformation import (
     AzureAIAnthropicCountTokensConfig,
@@ -69,6 +70,7 @@ class AzureAIAnthropicCountTokensHandler(AzureAIAnthropicCountTokensConfig):
                 tools=tools,
                 system=system,
                 optional_params=optional_params,
+                drop_params=drops_unsupported_params((litellm_params or {}).get("drop_params")),
             )
 
             verbose_logger.debug("Transformed request: %s", request_body)

litellm/llms/azure_ai/anthropic/handler.py

@@ -4,11 +4,12 @@ Azure Anthropic handler - reuses AnthropicChatCompletion logic with Azure authen
 
 import copy
 import json
-from collections.abc import Callable
+from collections.abc import Callable, Mapping
 from typing import Final
 
 import httpx
 
+from litellm.anthropic_beta_gated_fields import strip_beta_gated_request_body
 from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
 from litellm.llms.anthropic.chat.handler import AnthropicChatCompletion
 from litellm.llms.anthropic.chat.transformation import JsonModeRequest
@@ -75,13 +76,22 @@ class AzureAnthropicChatCompletion(AnthropicChatCompletion):
             litellm_params=litellm_params,
         )
 
-        data: Final = config.transform_request(
+        request_body: Final[Mapping[str, object]] = config.transform_request(
             model=model,
             messages=messages,
             optional_params=optional_params,
             litellm_params=litellm_params,
             headers=headers,
         )
+        data: Final = dict(
+            strip_beta_gated_request_body(
+                request_body,
+                headers,  # pyright: ignore[reportUnknownArgumentType]  # validate_environment returns a bare dict
+                drop_params=litellm_params.get("drop_params"),  # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType]  # litellm_params is the handler's bare dict
+                model=model,
+                provider=custom_llm_provider,
+            )
+        )
 
         ## LOGGING
         logging_obj.pre_call(

litellm/llms/azure_ai/anthropic/messages_transformation.py

@@ -145,6 +145,8 @@ class AzureAnthropicMessagesConfig(AnthropicMessagesConfig):
             litellm_params=litellm_params,
             headers=headers,
         )
-        self._normalize_system_role_messages(anthropic_messages_request, model=model)
+        self._normalize_system_role_messages(
+            anthropic_messages_request, model=model, drop_params=getattr(litellm_params, "drop_params", None)
+        )
         self._remove_scope_from_cache_control(anthropic_messages_request)
         return anthropic_messages_request

litellm/llms/azure_ai/anthropic/transformation.py

@@ -107,9 +107,11 @@ class AzureAnthropicConfig(AnthropicConfig):
             is_vertex_request=optional_params.get("is_vertex_request", False),
             user_anthropic_beta_headers=user_anthropic_beta_headers,
             mcp_server_used=mcp_server_used,
+            thinking=optional_params.get("thinking"),  # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType]  # optional_params is the base config's bare dict
         )
         # Merge headers - Azure auth (api-key or Authorization) takes precedence
-        headers = {**anthropic_headers, **headers}
+        merged_beta_header: Final = {k: v for k, v in anthropic_headers.items() if k == "anthropic-beta"}
+        headers = {**anthropic_headers, **headers, **merged_beta_header}
 
         # Ensure anthropic-version header is set
         if "anthropic-version" not in headers:

litellm/llms/azure_ai/chat/transformation.py

@@ -106,7 +106,7 @@ class AzureAIStudioConfig(OpenAIConfig):
         model: str,
         drop_params: bool,
         *,
-        declared_reasoning_efforts: Collection[str] = (),
+        declared_reasoning_efforts: Collection[str] | None = None,
     ) -> dict[str, object]:  # mutable-ok: OpenAIConfig.map_openai_params signature
         if not azureAIGPT5Config.is_model_gpt_5_model(model):
             return super().map_openai_params(

litellm/llms/azure_ai/passthrough/transformation.py

@@ -5,8 +5,9 @@ from types import MappingProxyType
 from typing import TYPE_CHECKING, Final
 
 import httpx
-from pydantic import TypeAdapter, ValidationError
+from pydantic import ConfigDict, TypeAdapter, ValidationError
 
+from litellm._lazy_imports import get_default_encoding
 from litellm._logging import verbose_logger
 from litellm.llms.azure_ai.common_utils import (
     AzureFoundryModelInfo,
@@ -22,10 +23,15 @@ from litellm.llms.base_llm.passthrough.transformation import (
     relayed_body,
     strip_leading_model_segment,
 )
+from litellm.llms.cohere.embed.transformation import (
+    CohereEmbeddingRowShape,
+    CohereImageTokenBilling,
+    cohere_embedding_response,
+)
 from litellm.rust_bridge.ocr.entrypoints import NATIVE_OCR_PASSTHROUGH_RESPONSE, NativeOcrPassthroughResponse
 from litellm.types.llms.openai import AllMessageValues
 from litellm.types.rerank import RerankResponse
-from litellm.types.utils import CallTypes, ImageResponse, StandardPassThroughResponseObject
+from litellm.types.utils import CallTypes, EmbeddingResponse, ImageResponse, StandardPassThroughResponseObject
 
 if TYPE_CHECKING:
     from httpx import URL, Response
@@ -35,6 +41,7 @@ if TYPE_CHECKING:
 
 
 EMPTY_QUERY: Final[Mapping[str, object]] = MappingProxyType({})
+ECHOED_TEXTS: Final = TypeAdapter(tuple[str, ...], config=ConfigDict(hide_input_in_errors=True))
 
 
 def api_version_from(litellm_params: Mapping[str, object]) -> str | None:
@@ -85,9 +92,29 @@ def relay_query_params(
     return MappingProxyType({**(request_query_params or EMPTY_QUERY), "api-version": api_version})
 
 
+def echoed_texts(body: Mapping[str, object]) -> tuple[str, ...]:
+    try:
+        return ECHOED_TEXTS.validate_python(body.get("texts"))
+    except ValidationError:
+        return ()
+
+
+def cohere_embed_relay_response(body: Mapping[str, object]) -> EmbeddingResponse:
+    return cohere_embedding_response(
+        body,
+        None,
+        get_default_encoding(),
+        echoed_texts(body),
+        CohereImageTokenBilling.FOLDED_INTO_INPUT_TOKENS,
+        CohereEmbeddingRowShape.TYPED,
+    )
+
+
 FOUNDRY_RELAY_SHAPES: Final = (
     RelayShape("/rerank", CallTypes.arerank, RerankResponse.model_validate),
     RelayShape("/providers/blackforestlabs/v1/flux-2-pro", CallTypes.aimage_generation, ImageResponse.model_validate),
+    RelayShape("/providers/cohere/v1/embed", CallTypes.aembedding, cohere_embed_relay_response),
+    RelayShape("/providers/cohere/v2/embed", CallTypes.aembedding, cohere_embed_relay_response),
 )
 
 

litellm/llms/azure_ai/responses/transformation.py

@@ -232,7 +232,7 @@ class AzureAIAgentResponsesAPIConfig(AzureAIResponsesAPIConfig):
         model: str,
         drop_params: bool,
         *,
-        declared_reasoning_efforts: Collection[str],
+        declared_reasoning_efforts: Collection[str] | None,
     ) -> Mapping[str, object]:
         mapped: Final = super().map_openai_params_for_deployment(
             response_api_optional_params, model, drop_params, declared_reasoning_efforts=declared_reasoning_efforts

litellm/llms/base_llm/base_utils.py

@@ -10,14 +10,23 @@ from abc import ABC, abstractmethod
 from collections.abc import Iterator, Mapping, Sequence
 from dataclasses import dataclass
 from types import MappingProxyType
-from typing import Any, Final
+from typing import Any, Final, TypeAlias
 
 from openai.lib import _parsing, _pydantic  # pyright: ignore[reportPrivateUsage]  # SDK parser internals
 from pydantic import BaseModel, JsonValue, TypeAdapter, ValidationError
+from typing_extensions import TypeIs  # noqa: TID251  # narrows a message to its system member by the role literal
 
 from litellm._logging import verbose_logger
+from litellm.constants import ANTHROPIC_BILLING_METADATA_PREFIX
 from litellm.litellm_core_utils.core_helpers import as_str_mapping
-from litellm.types.llms.openai import AllMessageValues, ChatCompletionToolCallChunk
+from litellm.types.llms.openai import (
+    AllMessageValues,
+    ChatCompletionCachedContent,
+    ChatCompletionSystemMessage,
+    ChatCompletionTextObject,
+    ChatCompletionToolCallChunk,
+    ChatCompletionUserMessage,
+)
 from litellm.types.utils import (
     CountContextManagement,
     Message,
@@ -458,7 +467,7 @@ def type_to_response_format_param(
 
 
 def map_developer_role_to_system_role(
-    messages: list[AllMessageValues],
+    messages: Sequence[AllMessageValues],
 ) -> list[AllMessageValues]:
     """
     Translate `developer` role to `system` role for non-OpenAI providers.
@@ -473,3 +482,226 @@ def map_developer_role_to_system_role(
         else:
             new_messages.append(m)
     return new_messages
+
+
+SystemMessageContent: TypeAlias = str | list[object]
+
+_system_content_adapter: Final = TypeAdapter[SystemMessageContent | None](SystemMessageContent | None)
+
+
+def _system_content(message: ChatCompletionSystemMessage) -> SystemMessageContent | None:
+    if "content" not in message:
+        return ""
+    try:
+        content: Final = _system_content_adapter.validate_python(message["content"])
+    except ValidationError:
+        return None
+    return "" if content is None else content
+
+
+def _message_cache_control(message: ChatCompletionSystemMessage) -> ChatCompletionCachedContent | None:
+    if "cache_control" not in message:
+        return None
+    return message["cache_control"]
+
+
+_SystemPart: TypeAlias = tuple[SystemMessageContent, ChatCompletionCachedContent | None]
+
+
+def _with_message_cache_control(
+    blocks: Sequence[object], cache_control: ChatCompletionCachedContent | None
+) -> tuple[object, ...]:
+    last: Final = as_str_mapping(blocks[-1]) if blocks else None
+    if last is None or cache_control is None or "cache_control" in last:
+        return tuple(blocks)
+    return (*blocks[:-1], {**last, "cache_control": cache_control})
+
+
+def _as_system_message(message: AllMessageValues) -> AllMessageValues:
+    if message["role"] != "developer":
+        return message
+    content_only: Final[ChatCompletionSystemMessage] = {"role": "system", "content": message["content"]}
+    named: Final[ChatCompletionSystemMessage] = (
+        content_only if "name" not in message else {**content_only, "name": message["name"]}
+    )
+    if "cache_control" not in message:
+        return named
+    cached: Final[ChatCompletionSystemMessage] = {**named, "cache_control": message["cache_control"]}
+    return cached
+
+
+def _text_blocks(
+    content: SystemMessageContent, cache_control: ChatCompletionCachedContent | None
+) -> tuple[object, ...]:
+    if not isinstance(content, str):
+        return _with_message_cache_control(content, cache_control)
+    if not content:
+        return ()
+    if cache_control:
+        cached_block: Final[ChatCompletionTextObject] = {
+            "type": "text",
+            "text": content,
+            "cache_control": cache_control,
+        }
+        return (cached_block,)
+    text_block: Final[ChatCompletionTextObject] = {"type": "text", "text": content}
+    return (text_block,)
+
+
+def _lone_text_block_text(blocks: Sequence[object]) -> str | None:
+    match blocks:
+        case [{"type": "text", "text": str() as text} as block] if block == {"type": "text", "text": text}:
+            return text
+        case _:
+            return None
+
+
+def _plain_text(content: SystemMessageContent, cache_control: ChatCompletionCachedContent | None) -> str | None:
+    if cache_control:
+        return None
+    text: Final = content if isinstance(content, str) else _lone_text_block_text(content)
+    if text is None or text.startswith(ANTHROPIC_BILLING_METADATA_PREFIX):
+        return None
+    return text
+
+
+def _is_system_message(
+    message: AllMessageValues,
+) -> TypeIs[ChatCompletionSystemMessage]:  # guard-ok: the role literal picks the TypedDict member
+    return message["role"] == "system"
+
+
+def _with_leading_separator(blocks: tuple[object, ...]) -> tuple[object, ...]:
+    first: Final = as_str_mapping(blocks[0])
+    text: Final = None if first is None else first.get("text")
+    if first is None or not isinstance(text, str) or text.startswith(ANTHROPIC_BILLING_METADATA_PREFIX):
+        return blocks
+    return ({**first, "text": f"\n\n{text}"}, *blocks[1:])
+
+
+def _merged_system_content(parts: Sequence[_SystemPart]) -> SystemMessageContent:
+    texts: Final = tuple(itertools.starmap(_plain_text, parts))
+    if all(text is not None for text in texts):
+        return "\n\n".join(text for text in texts if text)
+    block_runs: Final = tuple(blocks for blocks in itertools.starmap(_text_blocks, parts) if blocks)
+    separated: Final = (*block_runs[:1], *map(_with_leading_separator, block_runs[1:]))
+    return list(itertools.chain.from_iterable(separated))
+
+
+def _merged_system_message(
+    run: Sequence[ChatCompletionSystemMessage], parts: Sequence[_SystemPart]
+) -> ChatCompletionSystemMessage:
+    content: Final = _merged_system_content(parts)
+    named: Final = next((message for message in reversed(run) if "name" in message), None)
+    if named is None:
+        merged: Final[ChatCompletionSystemMessage] = {"role": "system", "content": content}
+        return merged
+    with_name: Final[ChatCompletionSystemMessage] = {"role": "system", "content": content, "name": named["name"]}
+    return with_name
+
+
+def _merged_run(run: tuple[AllMessageValues, ...]) -> tuple[AllMessageValues, ...]:
+    system_run: Final = tuple(message for message in run if _is_system_message(message))
+    if len(system_run) < 2:
+        return run
+    contents: Final = tuple((_system_content(message), _message_cache_control(message)) for message in system_run)
+    parts: Final = tuple((content, cache_control) for content, cache_control in contents if content is not None)
+    if len(parts) != len(contents):
+        return run
+    return (_merged_system_message(system_run, parts),)
+
+
+def _merged_system_runs(messages: Sequence[AllMessageValues]) -> Iterator[AllMessageValues]:
+    runs: Final = (tuple(run) for _, run in itertools.groupby(messages, key=_is_system_message))
+    return itertools.chain.from_iterable(map(_merged_run, runs))
+
+
+_INSTRUCTION_ROLES: Final = frozenset(("system", "developer"))
+
+
+def _leading_instruction_block_length(roles: Sequence[str | None]) -> int:
+    return next((index for index, role in enumerate(roles) if role not in _INSTRUCTION_ROLES), len(roles))
+
+
+def _closing_instruction_block_start(roles: Sequence[str | None], leading_length: int) -> int:
+    last_conversation_index: Final = next(
+        (index for index in range(len(roles) - 1, leading_length - 1, -1) if roles[index] not in _INSTRUCTION_ROLES),
+        None,
+    )
+    if last_conversation_index is None or roles[last_conversation_index] != "assistant":
+        return len(roles)
+    return last_conversation_index + 1
+
+
+def hoisted_instruction_indices(roles: Sequence[str | None]) -> tuple[int, ...]:
+    """
+    Indices of the leading instruction block plus every developer item after it, except
+    a developer block that closes the conversation right after an assistant turn, which
+    stays in place so the request does not end on the assistant's turn. Items without a
+    role (tool calls and their outputs) count as conversation.
+    """
+    leading_length: Final = _leading_instruction_block_length(roles)
+    closing_start: Final = _closing_instruction_block_start(roles, leading_length)
+    moved: Final = tuple(index for index in range(leading_length, closing_start) if roles[index] == "developer")
+    return (*range(leading_length), *moved)
+
+
+def closing_instruction_indices(roles: Sequence[str | None]) -> frozenset[int]:
+    closing_start: Final = _closing_instruction_block_start(roles, _leading_instruction_block_length(roles))
+    return frozenset(range(closing_start, len(roles)))
+
+
+def _role_name(role: object) -> str | None:
+    return role if isinstance(role, str) else None
+
+
+def _message_roles(messages: Sequence[AllMessageValues]) -> tuple[str | None, ...]:
+    return tuple(_role_name(message["role"]) for message in messages)
+
+
+def _hoisted_message_order(messages: Sequence[AllMessageValues]) -> tuple[int, ...]:
+    hoisted: Final = hoisted_instruction_indices(_message_roles(messages))
+    hoisted_set: Final = frozenset(hoisted)
+    return (*hoisted, *(index for index in range(len(messages)) if index not in hoisted_set))
+
+
+def _developer_message_as_user(message: AllMessageValues) -> AllMessageValues:
+    if message["role"] != "developer":
+        return message
+    copied: Final = copy.deepcopy(message)
+    if "cache_control" not in copied:
+        as_user: Final[ChatCompletionUserMessage] = {"role": "user", "content": copied["content"]}
+        return as_user
+    cached_as_user: Final[ChatCompletionUserMessage] = {
+        "role": "user",
+        "content": copied["content"],
+        "cache_control": copied["cache_control"],
+    }
+    return cached_as_user
+
+
+def merge_system_message_runs(messages: Sequence[AllMessageValues]) -> Sequence[AllMessageValues]:
+    """
+    Translate `developer` role to `system` role in place and fold each run of
+    consecutive system messages into one, for chat templates that render a single
+    system message per run.
+    """
+    return tuple(_merged_system_runs(tuple(map(_as_system_message, messages))))
+
+
+def hoist_developer_messages_into_leading_system_message(
+    messages: Sequence[AllMessageValues],
+) -> Sequence[AllMessageValues]:
+    """
+    Place instructions for an OpenAI-compatible model whose chat template allows a single
+    system message and only at the start: developer messages after the first user turn move
+    into the leading system block, a developer message that closes the conversation right
+    after an assistant turn becomes a user turn so the request does not end on the
+    assistant's turn, and each run of consecutive system messages folds into one.
+    """
+    order: Final = _hoisted_message_order(messages)
+    closing: Final = closing_instruction_indices(_message_roles(messages))
+    placed: Final = tuple(
+        _developer_message_as_user(messages[index]) if index in closing else messages[index] for index in order
+    )
+    return merge_system_message_runs(placed)

litellm/llms/base_llm/chat/transformation.py

@@ -4,7 +4,7 @@ Common base config for all LLM providers
 
 import types
 from abc import ABC, abstractmethod
-from collections.abc import AsyncIterator, Iterator, Mapping
+from collections.abc import AsyncIterator, Iterator, Mapping, Sequence
 from typing import TYPE_CHECKING, Any, Final, Union
 
 import httpx
@@ -199,12 +199,19 @@ class BaseConfig(ABC):
 
     def translate_developer_role_to_system_role(
         self,
-        messages: list[AllMessageValues],
-    ) -> list[AllMessageValues]:
+        messages: Sequence[AllMessageValues],
+        *,
+        model: str,
+        custom_llm_provider: str | None,
+        api_base: str | None,
+        model_info: object,
+    ) -> Sequence[AllMessageValues]:
         """
         Translate `developer` role to `system` role for non-OpenAI providers.
 
-        Overriden by OpenAI/Azure
+        Overridden by OpenAI, which also places the instructions in one leading system
+        message for an OpenAI-compatible model declared to reject a later system
+        message, and by the o-series and Azure configs.
         """
         return map_developer_role_to_system_role(messages=messages)
 

litellm/llms/base_llm/responses/transformation.py

@@ -69,6 +69,9 @@ class BaseResponsesAPIConfig(ABC):
     def supports_encrypted_agent_messages(self) -> bool:
         return False
 
+    def supports_streamed_retrieval(self) -> bool:
+        return self.custom_llm_provider in (LlmProviders.OPENAI, LlmProviders.AZURE, LlmProviders.AZURE_AI)
+
     def sign_request(
         self,
         headers: dict,
@@ -209,7 +212,7 @@ class BaseResponsesAPIConfig(ABC):
         api_base: str,
         litellm_params: GenericLiteLLMParams,
         headers: dict,
-    ) -> tuple[str, dict]:
+    ) -> tuple[str, dict[str, str]]:
         pass
 
     @abstractmethod

litellm/llms/bedrock/chat/converse_transformation.py

@@ -17,6 +17,8 @@ from typing_extensions import ReadOnly, TypedDict
 
 import litellm
 from litellm._logging import verbose_logger
+from litellm.anthropic_beta_gated_fields import active_beta_values, log_beta_gated_strips, strip_beta_gated_thinking
+from litellm.anthropic_beta_headers_manager import filter_and_transform_beta_headers
 from litellm.constants import (
     BEDROCK_MIN_THINKING_BUDGET_TOKENS,
     RESPONSE_FORMAT_TOOL_NAME,
@@ -117,6 +119,7 @@ from litellm.utils import (
 )
 
 from ..common_utils import (
+    DROP_DISABLED_REASONING_EFFORT_WARNING,
     BedrockError,
     BedrockModelInfo,
     bedrock_cache_point_ttl_supported,
@@ -1192,11 +1195,7 @@ class AmazonConverseConfig(BaseConfig):
                         ),
                         status_code=400,
                     )
-                verbose_logger.debug(
-                    "Dropping unsupported `reasoning_effort=%s` for Bedrock model=%s.",
-                    value,
-                    model,
-                )
+                verbose_logger.warning(DROP_DISABLED_REASONING_EFFORT_WARNING, "reasoning_effort", value, model)
             elif param == "reasoning_effort" and isinstance(value, str):
                 self._handle_reasoning_effort_parameter(
                     model=model, reasoning_effort=value, optional_params=optional_params
@@ -1828,7 +1827,16 @@ class AmazonConverseConfig(BaseConfig):
                 filtered_tools.append(tool)
 
         base_model: Final = BedrockModelInfo.get_base_model(model)
-        client_beta_list: Final = get_anthropic_beta_from_headers(headers or {})
+        header_beta_list: Final = get_anthropic_beta_from_headers(headers or {})
+        client_beta_list: Final = header_beta_list or filter_and_transform_beta_headers(
+            sorted(
+                active_beta_values(
+                    {},
+                    additional_request_params.get("anthropic_beta"),  # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType]  # additional_request_params is a bare dict
+                )
+            ),
+            provider="bedrock_converse",
+        )
         eager_beta: Final = (
             (ANTHROPIC_FINE_GRAINED_TOOL_STREAMING_BETA_HEADER,)
             if base_model.startswith("anthropic")
@@ -1836,7 +1844,15 @@ class AmazonConverseConfig(BaseConfig):
             and ANTHROPIC_FINE_GRAINED_TOOL_STREAMING_BETA_HEADER not in client_beta_list
             else ()
         )
-        anthropic_beta_list: Final = [*client_beta_list, *eager_beta]
+        thinking_betas: Final = AnthropicModelInfo().thinking_betas(
+            additional_request_params.get("thinking"),  # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType]  # additional_request_params is a bare dict
+        )
+        missing_thinking_betas: Final = (
+            tuple(beta for beta in thinking_betas if beta not in client_beta_list)
+            if base_model.startswith("anthropic")
+            else ()
+        )
+        anthropic_beta_list: Final = [*client_beta_list, *eager_beta, *missing_thinking_betas]
 
         # Only separate tools if computer use tools are actually present
         if filtered_tools and self.is_computer_use_tool_used(filtered_tools, model):
@@ -1906,7 +1922,8 @@ class AmazonConverseConfig(BaseConfig):
                 else:
                     computer_use_header = "computer-use-2024-10-22"
 
-                anthropic_beta_list.append(computer_use_header)
+                if computer_use_header not in anthropic_beta_list:
+                    anthropic_beta_list.append(computer_use_header)
                 # Transform computer use tools to proper Bedrock format
                 transformed_computer_tools: Final = self._transform_computer_use_tools(computer_use_tools)
                 additional_request_params["tools"] = transformed_computer_tools
@@ -2031,6 +2048,16 @@ class AmazonConverseConfig(BaseConfig):
         bedrock_tools, anthropic_beta_list = self._process_tools_and_beta(
             original_tools, model, headers, additional_request_params, litellm_params, deployment_model
         )
+        requested_thinking: Final = additional_request_params.get("thinking")  # pyright: ignore[reportUnknownMemberType, reportUnknownVariableType]  # _prepare_request_params hands back a bare dict
+        body_betas: Final = additional_request_params.get("anthropic_beta")  # pyright: ignore[reportUnknownMemberType, reportUnknownVariableType]  # same bare dict
+        gated_thinking: Final = strip_beta_gated_thinking(
+            requested_thinking,  # pyright: ignore[reportUnknownArgumentType]  # same bare dict
+            active_beta_values({}, body_betas) | frozenset(anthropic_beta_list),  # pyright: ignore[reportUnknownArgumentType]  # _process_tools_and_beta hands back a bare list
+            drop_params=drop_params,
+        )
+        if gated_thinking.stripped:
+            log_beta_gated_strips(gated_thinking.stripped, model=model, provider="bedrock_converse")
+            additional_request_params["thinking"] = gated_thinking.thinking
 
         # Append cachePoint to tools if cache_control_injection_points has tool_config
         cache_injection_points: Final = additional_request_params.pop("cache_control_injection_points", None)

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

@@ -4,6 +4,7 @@ from typing import TYPE_CHECKING, Any, Final
 import httpx
 from pydantic import TypeAdapter
 
+from litellm.anthropic_beta_gated_fields import strip_beta_gated_request_body
 from litellm.anthropic_beta_headers_manager import filter_and_transform_beta_headers
 from litellm.litellm_core_utils.llm_cost_calc.guardrail_cost import object_mapping
 from litellm.litellm_core_utils.prompt_templates.factory import (
@@ -210,7 +211,15 @@ class AmazonAnthropicClaudeConfig(AmazonInvokeConfig, AnthropicConfig):
         if beta_list:
             _anthropic_request["anthropic_beta"] = beta_list
 
-        return _anthropic_request
+        return dict(
+            strip_beta_gated_request_body(
+                _anthropic_request,  # pyright: ignore[reportUnknownArgumentType]  # built by the base signature's bare dict transform
+                {},
+                drop_params=litellm_params.get("drop_params"),  # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType]  # litellm_params is the base signature's bare dict
+                model=model,
+                provider="bedrock_invoke",
+            )
+        )
 
     @property
     def uses_async_transform_request(self) -> bool:
@@ -294,7 +303,6 @@ class AmazonAnthropicClaudeConfig(AmazonInvokeConfig, AnthropicConfig):
         programmatic_tool_calling_used: Final = self.is_programmatic_tool_calling_used(tools)
         input_examples_used: Final = self.is_input_examples_used(tools)
         is_mid_conversation_output_config_used: Final = self.is_mid_conversation_output_config_used(messages)
-        is_thinking_display_updates_used: Final = self.is_thinking_display_updates_used(thinking)
 
         client_beta_set: Final = set(get_anthropic_beta_from_headers(headers))
         user_beta_set: Final = (
@@ -312,7 +320,7 @@ class AmazonAnthropicClaudeConfig(AmazonInvokeConfig, AnthropicConfig):
             mcp_server_used=self.is_mcp_server_used(optional_params.get("mcp_servers")),
             custom_llm_provider="bedrock",
             is_mid_conversation_output_config_used=is_mid_conversation_output_config_used,
-            is_thinking_display_updates_used=is_thinking_display_updates_used,
+            thinking=thinking,
         )
         beta_set.update(auto_betas)
 

litellm/llms/bedrock/claude_platform/transformation.py

@@ -72,6 +72,7 @@ class BedrockClaudePlatformConfig(BedrockClaudePlatformMixin, AnthropicConfig):
             ),
             code_execution_tool_used=self.is_code_execution_tool_used(tools=optional_params.get("tools")),
             container_with_skills_used=self.is_container_with_skills_used(optional_params=optional_params),
+            thinking=optional_params.get("thinking"),  # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType]  # optional_params is the base config's bare dict
         )
         anthropic_headers["anthropic-workspace-id"] = workspace_id
         return {**headers, **anthropic_headers}

litellm/llms/bedrock/common_utils.py

@@ -15,7 +15,11 @@ from dataclasses import dataclass
 from types import MappingProxyType
 from typing import TYPE_CHECKING, Any, Final, Literal, TypeAlias
 
-from typing_extensions import ReadOnly, TypedDict
+from typing_extensions import (
+    ReadOnly,
+    TypedDict,
+    TypeIs,  # noqa: TID251  # narrows the caller's tools without copying them
+)
 
 if TYPE_CHECKING:
     from botocore.eventstream import EventStreamMessage
@@ -573,6 +577,26 @@ def strip_unsupported_bedrock_invoke_output_config_keys(
         request_body["output_config"] = {"format": preserved_format}  # rebind-ok: out-param
 
 
+_TOOL_ITEMS_ADAPTER: Final = TypeAdapter(tuple[object, ...])
+
+
+def _is_str_keyed_dict(
+    value: object,
+) -> TypeIs[dict[str, object]]:  # guard-ok: JSON tool objects  # mutable-ok: fallbacks reuse tools edited in place
+    return isinstance(value, dict)
+
+
+def _normalize_custom_field_on_tool(tool: object) -> None:
+    if not _is_str_keyed_dict(tool):
+        return
+    custom: Final = tool.pop("custom", None)
+    if not _is_str_keyed_dict(custom) or "defer_loading" in tool:
+        return
+    deferred: Final = custom.get("defer_loading")
+    if isinstance(deferred, bool):
+        tool["defer_loading"] = deferred  # rebind-ok: a fallback reuses the caller's tool, as on main
+
+
 def normalize_custom_field_on_tools(request_body: dict) -> None:
     """
     Drop the ``custom`` field from each tool, first hoisting a boolean
@@ -590,15 +614,8 @@ def normalize_custom_field_on_tools(request_body: dict) -> None:
     tools: Final = request_body.get("tools")
     if not tools or not isinstance(tools, list):
         return
-    for tool in tools:
-        if not isinstance(tool, dict):
-            continue
-        custom: dict[str, object] | None = tool.pop("custom", None)
-        if not isinstance(custom, dict) or "defer_loading" in tool:
-            continue
-        deferred: object = custom.get("defer_loading")
-        if isinstance(deferred, bool):
-            tool["defer_loading"] = deferred
+    for tool in _TOOL_ITEMS_ADAPTER.validate_python(tools):
+        _normalize_custom_field_on_tool(tool)
 
 
 _TOOL_DICTS_ADAPTER: Final = TypeAdapter(tuple[Mapping[str, object], ...])
@@ -1314,6 +1331,12 @@ def _mantle_api_base_from_env() -> str | None:
     return next((base[: -len(suffix)] for suffix in _MANTLE_OPENAI_BASE_SUFFIXES if base.endswith(suffix)), base)
 
 
+DROP_DISABLED_REASONING_EFFORT_WARNING: Final = (
+    "Dropping %s=%r for Bedrock model=%r (drop_params=True) because the model map marks that level unsupported. "
+    "Send a level the model supports."
+)
+
+
 def bedrock_reasoning_effort_disabled(model: str, effort: str) -> bool:
     from litellm.utils import is_explicitly_disabled_factory
 

litellm/llms/bedrock/count_tokens/mantle_handler.py

@@ -8,6 +8,7 @@ from typing_extensions import NotRequired, ReadOnly, TypedDict
 import litellm
 from litellm._logging import verbose_logger
 from litellm.litellm_core_utils.url_utils import join_api_base
+from litellm.llms.anthropic.common_utils import tools_without_custom_field
 from litellm.llms.anthropic.count_tokens.transformation import COUNT_TOKEN_OPTION_NAMES, AnthropicCountTokensConfig
 from litellm.llms.bedrock.base_aws_llm import BaseAWSLLM, run_aws_signing
 from litellm.llms.bedrock.common_utils import BedrockError, build_mantle_messages_url
@@ -26,6 +27,7 @@ class MantleCountTokensRequest(TypedDict):
 _COUNT_REQUEST: Final = TypeAdapter(MantleCountTokensRequest)
 _COUNT_RESPONSE: Final = TypeAdapter(dict[str, JsonValue])
 _COUNT_OPTIONS: Final[TypeAdapter[Mapping[str, JsonValue]]] = TypeAdapter(Mapping[str, JsonValue])
+_TOOLS: Final = TypeAdapter(list[dict[str, JsonValue]])
 
 
 class BedrockMantleCountTokensHandler(BaseAWSLLM):
@@ -59,10 +61,15 @@ class BedrockMantleCountTokensHandler(BaseAWSLLM):
                 self.get_aws_region_name, optional_params=litellm_params, model=resolved_model, model_id=None
             )
             endpoint_url: Final = await run_aws_signing(self.get_mantle_count_tokens_endpoint, aws_region_name)
+            tools: Final = request.get("tools")
             body: Final = self._anthropic_config.transform_request_to_count_tokens(
                 model=resolved_model,
                 messages=request["messages"],
-                tools=request.get("tools"),
+                tools=None
+                if tools is None
+                else _TOOLS.validate_python(
+                    tools_without_custom_field(tools, hoist_defer_loading=True, drop_custom=True)
+                ),
                 system=request.get("system"),
                 optional_params=_COUNT_OPTIONS.validate_python(
                     {name: value for name, value in request_data.items() if name in COUNT_TOKEN_OPTION_NAMES}

litellm/llms/bedrock/count_tokens/transformation.py

@@ -12,6 +12,7 @@ from typing import Final, Literal
 from pydantic import JsonValue
 
 from litellm.litellm_core_utils.url_utils import join_api_base
+from litellm.llms.anthropic.common_utils import tools_without_custom_field
 from litellm.llms.bedrock.base_aws_llm import BaseAWSLLM
 from litellm.llms.bedrock.common_utils import get_bedrock_base_model
 
@@ -194,7 +195,15 @@ class BedrockCountTokensConfig(BaseAWSLLM):
 
         # For InvokeModel, we need to provide the raw body that would be sent to the model
         # Remove the 'model' field from the body as it's not part of the model input
-        body_data: Final = {k: v for k, v in request_data.items() if k != "model"}
+        tools: Final = request_data.get("tools")
+        body_data: Final = {
+            **{k: v for k, v in request_data.items() if k != "model"},
+            **(
+                {"tools": tools_without_custom_field(tools, hoist_defer_loading=True, drop_custom=True)}
+                if isinstance(tools, list)
+                else {}
+            ),
+        }
 
         if "messages" in body_data:
             # Bedrock validates the body against the model's InvokeModel schema;

litellm/llms/bedrock/embed/embedding.py

@@ -9,9 +9,11 @@ from collections.abc import Callable, Mapping
 from typing import TYPE_CHECKING, Final, get_args, overload
 
 import httpx
+from pydantic import TypeAdapter
 
 import litellm
 from litellm.constants import BEDROCK_EMBEDDING_PROVIDERS_LITERAL
+from litellm.litellm_core_utils.hidden_params import set_hidden_param
 from litellm.litellm_core_utils.url_utils import join_api_base
 from litellm.llms.cohere.embed.handler import embedding as cohere_embedding
 from litellm.llms.custom_httpx.http_handler import (
@@ -36,7 +38,7 @@ from ..base_aws_llm import (
     pop_aws_auth_params,
     run_aws_signing,
 )
-from ..common_utils import BedrockError
+from ..common_utils import BedrockError, split_bedrock_region_path
 from .amazon_nova_transformation import AmazonNovaEmbeddingConfig
 from .amazon_titan_g1_transformation import AmazonTitanG1Config
 from .amazon_titan_multimodal_transformation import (
@@ -64,12 +66,21 @@ def _sign_get_request(
     return request.prepare()
 
 
+_OPTIONAL_STR: Final[TypeAdapter[str | None]] = TypeAdapter(str | None)
+
+
+def _without_region_path(model: str) -> str:
+    region, model_id = split_bedrock_region_path(model)
+    return model if region is None else model_id
+
+
 class BedrockEmbedding(BaseAWSLLM):
     @overload
     def _load_credentials(
         self,
         optional_params: dict,  # mutable-ok: the implementation pops the aws_* keys out of the caller's dict in place
         bearer_token: None = None,
+        region_from_model: str | None = None,
     ) -> tuple[Credentials, str]: ...
 
     @overload
@@ -77,15 +88,18 @@ class BedrockEmbedding(BaseAWSLLM):
         self,
         optional_params: dict,  # mutable-ok: the implementation pops the aws_* keys out of the caller's dict in place
         bearer_token: str,
+        region_from_model: str | None = None,
     ) -> tuple[None, str]: ...
 
     def _load_credentials(
         self,
         optional_params: dict,
         bearer_token: str | None = None,
+        region_from_model: str | None = None,
     ) -> tuple[Credentials | None, str]:
         auth_params: Final = pop_aws_auth_params(optional_params)
-        aws_region_name = optional_params.pop("aws_region_name", None)
+        explicit_region_name: Final = optional_params.pop("aws_region_name", None)
+        aws_region_name = region_from_model if explicit_region_name is None else explicit_region_name
 
         ### SET REGION NAME ###
         if aws_region_name is None:
@@ -184,14 +198,17 @@ class BedrockEmbedding(BaseAWSLLM):
         provider: BEDROCK_EMBEDDING_PROVIDERS_LITERAL,
         is_async_invoke: bool | None = False,
         batch_data: list[dict] | None = None,
+        aws_region_name: str | None = None,
     ) -> EmbeddingResponse | None:
         """
         Transforms the response from the Bedrock embedding provider to the OpenAI format.
         """
         returned_response: EmbeddingResponse | None = None
+        bedrock_model: Final = _without_region_path(model)
+        is_async_job: Final = bool(is_async_invoke) and len(response_list) == 1 and "invocationArn" in response_list[0]
 
         # Handle async invoke responses (single response with invocationArn)
-        if is_async_invoke and len(response_list) == 1 and "invocationArn" in response_list[0]:
+        if is_async_job:
             if provider == "twelvelabs":
                 returned_response = TwelveLabsMarengoEmbeddingConfig().transform_async_invoke_response(
                     response=response_list[0], model=model
@@ -227,15 +244,15 @@ class BedrockEmbedding(BaseAWSLLM):
                 )
         else:
             # Handle regular invoke responses
-            if model == "amazon.titan-embed-image-v1":
+            if bedrock_model == "amazon.titan-embed-image-v1":
                 returned_response = AmazonTitanMultimodalEmbeddingG1Config().transform_response(
                     response_list=response_list, model=model, batch_data=batch_data
                 )
-            elif model == "amazon.titan-embed-text-v1":
+            elif bedrock_model == "amazon.titan-embed-text-v1":
                 returned_response = AmazonTitanG1Config().transform_response(response_list=response_list, model=model)
-            elif model == "amazon.titan-embed-text-v2:0":
+            elif bedrock_model == "amazon.titan-embed-text-v2:0":
                 returned_response = AmazonTitanV2Config().transform_response(response_list=response_list, model=model)
-            elif model == "amazon.titan-embed-g1-text-02":
+            elif bedrock_model == "amazon.titan-embed-g1-text-02":
                 returned_response = AmazonTitanG1Config().transform_response(response_list=response_list, model=model)
             elif provider == "twelvelabs":
                 returned_response = TwelveLabsMarengoEmbeddingConfig().transform_response(
@@ -251,6 +268,8 @@ class BedrockEmbedding(BaseAWSLLM):
         ##########################################################
         if returned_response is None:
             raise Exception(f"Unable to map model response to known provider format. model={model}")
+        if aws_region_name is not None and not is_async_job:
+            set_hidden_param(returned_response, "region_name", aws_region_name)
         return returned_response
 
     def _single_func_embeddings(
@@ -319,6 +338,7 @@ class BedrockEmbedding(BaseAWSLLM):
             provider=provider,
             is_async_invoke=is_async_invoke,
             batch_data=batch_data,
+            aws_region_name=aws_region_name,
         )
 
     async def _async_single_func_embeddings(
@@ -390,6 +410,7 @@ class BedrockEmbedding(BaseAWSLLM):
             provider=provider,
             is_async_invoke=is_async_invoke,
             batch_data=batch_data,
+            aws_region_name=aws_region_name,
         )
 
     def embeddings(
@@ -409,26 +430,29 @@ class BedrockEmbedding(BaseAWSLLM):
         litellm_params: dict,
         api_key: str | None = None,
     ) -> EmbeddingResponse:
+        model_id_param: Final = _OPTIONAL_STR.validate_python(optional_params.pop("model_id", None)) or None
+        has_async_invoke: Final = "async_invoke/" in model
+        model_without_async_route: Final = model.replace("async_invoke/", "", 1)
+        path_region: Final = split_bedrock_region_path(model_without_async_route)[0] if model_id_param is None else None
+        bedrock_model: Final = _without_region_path(model_without_async_route)
         credentials, aws_region_name = self._load_credentials(
-            optional_params, bearer_token=bedrock_bearer_token(api_key)
+            optional_params,
+            bearer_token=bedrock_bearer_token(api_key),
+            region_from_model=self._get_aws_region_from_model_arn(model_id_param) or path_region,
         )
 
         ### TRANSFORMATION ###
-        unencoded_model_id: Final = optional_params.pop("model_id", None) or model  # default to model if not passed
+        unencoded_model_id: Final = model_id_param or bedrock_model
         modelId: Final = urllib.parse.quote(unencoded_model_id, safe="")
         aws_region_name = self._get_aws_region_name(
             optional_params={"aws_region_name": aws_region_name},
-            model=model,
+            model=bedrock_model,
             model_id=unencoded_model_id,
         )
-        # Check async invoke needs to be used
-        has_async_invoke: Final = "async_invoke/" in model
-        if has_async_invoke:
-            model = model.replace("async_invoke/", "", 1)
-        provider: Final = self.get_bedrock_embedding_provider(model)
+        provider: Final = self.get_bedrock_embedding_provider(bedrock_model)
         if provider is None:
             raise Exception(
-                f"Unable to determine bedrock embedding provider for model: {model}. "
+                f"Unable to determine bedrock embedding provider for model: {bedrock_model}. "
                 f"Supported providers: {list(get_args(BEDROCK_EMBEDDING_PROVIDERS_LITERAL))}"
             )
         inference_params = copy.deepcopy(optional_params)
@@ -441,9 +465,9 @@ class BedrockEmbedding(BaseAWSLLM):
         batch_data: list | None = None
         if provider == "cohere":
             data = BedrockCohereEmbeddingConfig().transform_request(
-                model=model, input=input, inference_params=inference_params
+                model=bedrock_model, input=input, inference_params=inference_params
             )
-        elif provider == "amazon" and model in [
+        elif provider == "amazon" and bedrock_model in [
             "amazon.titan-embed-image-v1",
             "amazon.titan-embed-text-v1",
             "amazon.titan-embed-text-v2:0",
@@ -451,28 +475,28 @@ class BedrockEmbedding(BaseAWSLLM):
         ]:
             batch_data = []
             for i in input:
-                if model == "amazon.titan-embed-image-v1":
+                if bedrock_model == "amazon.titan-embed-image-v1":
                     transformed_request: AmazonEmbeddingRequest = (
                         AmazonTitanMultimodalEmbeddingG1Config().transform_request(
                             input=i, inference_params=inference_params
                         )
                     )
-                elif model == "amazon.titan-embed-text-v1":
+                elif bedrock_model == "amazon.titan-embed-text-v1":
                     transformed_request = AmazonTitanG1Config().transform_request(
                         input=i, inference_params=inference_params
                     )
-                elif model == "amazon.titan-embed-text-v2:0":
+                elif bedrock_model == "amazon.titan-embed-text-v2:0":
                     transformed_request = AmazonTitanV2Config().transform_request(
                         input=i, inference_params=inference_params
                     )
-                elif model == "amazon.titan-embed-g1-text-02":
+                elif bedrock_model == "amazon.titan-embed-g1-text-02":
                     transformed_request = AmazonTitanG1Config().transform_request(
                         input=i, inference_params=inference_params
                     )
                 else:
                     raise Exception(
                         "Unmapped model. Received={}. Expected={}".format(
-                            model,
+                            bedrock_model,
                             [
                                 "amazon.titan-embed-image-v1",
                                 "amazon.titan-embed-text-v1",
@@ -485,7 +509,7 @@ class BedrockEmbedding(BaseAWSLLM):
         elif provider == "twelvelabs":
             batch_data = []
             for i in input:
-                twelvelabs_request = TwelveLabsMarengoEmbeddingConfig(model=model).transform_request(
+                twelvelabs_request = TwelveLabsMarengoEmbeddingConfig(model=bedrock_model).transform_request(
                     input=i,
                     inference_params=inference_params,
                     async_invoke_route=has_async_invoke,
@@ -527,7 +551,7 @@ class BedrockEmbedding(BaseAWSLLM):
                     extra_headers=extra_headers,
                     endpoint_url=endpoint_url,
                     aws_region_name=aws_region_name,
-                    model=model,
+                    model=model_without_async_route,
                     logging_obj=logging_obj,
                     api_key=api_key,
                     provider=provider,
@@ -541,7 +565,7 @@ class BedrockEmbedding(BaseAWSLLM):
                 extra_headers=extra_headers,
                 endpoint_url=endpoint_url,
                 aws_region_name=aws_region_name,
-                model=model,
+                model=model_without_async_route,
                 logging_obj=logging_obj,
                 api_key=api_key,
                 provider=provider,
@@ -570,8 +594,9 @@ class BedrockEmbedding(BaseAWSLLM):
         ## ROUTING ##
         # Convert CaseInsensitiveDict to regular dict for httpx compatibility
         headers_for_request: Final = dict(prepped.headers) if hasattr(prepped, "headers") else {}
+        set_hidden_param(model_response, "region_name", aws_region_name)
         return cohere_embedding(
-            model=model,
+            model=model_without_async_route,
             input=input,
             model_response=model_response,
             logging_obj=logging_obj,

litellm/llms/bedrock/messages/invoke_transformations/anthropic_claude_transformation.py

@@ -7,6 +7,7 @@ from pydantic import TypeAdapter
 
 import litellm
 from litellm._logging import verbose_logger
+from litellm.anthropic_beta_gated_fields import log_beta_gated_strips, strip_beta_gated_native_request
 from litellm.anthropic_beta_headers_manager import filter_and_transform_beta_headers
 from litellm.constants import (
     BEDROCK_MIN_THINKING_BUDGET_TOKENS,
@@ -102,6 +103,9 @@ class AmazonAnthropicClaudeMessagesConfig(
     def should_strip_billing_metadata(self) -> bool:
         return True
 
+    def tools_for_upstream(self, tools: Sequence[object], model: str, drop_params: object) -> Sequence[object]:
+        return tools
+
     def drops_dangerous_tool_use_beta_without_safeguards(self) -> bool:
         return True
 
@@ -546,6 +550,9 @@ class AmazonAnthropicClaudeMessagesConfig(
             "edits": retained_edits,
         }
 
+    def strips_beta_gated_fields_from_final_headers(self) -> bool:
+        return False
+
     def _get_bedrock_invoke_anthropic_beta_headers(
         self,
         model: str,
@@ -581,9 +588,7 @@ class AmazonAnthropicClaudeMessagesConfig(
             ),
             custom_llm_provider="bedrock",
             is_mid_conversation_output_config_used=is_mid_conversation_output_config_used,
-            is_thinking_display_updates_used=anthropic_model_info.is_thinking_display_updates_used(
-                anthropic_messages_request.get("thinking")
-            ),
+            thinking=anthropic_messages_request.get("thinking"),
             is_mid_conversation_tool_change_used=anthropic_model_info.is_mid_conversation_tool_change_used(
                 outgoing_messages_typed
             ),
@@ -754,7 +759,9 @@ class AmazonAnthropicClaudeMessagesConfig(
                 ),
                 "display": "updates",
             }
-        self._normalize_system_role_messages(anthropic_messages_request, model=model)
+        self._normalize_system_role_messages(
+            anthropic_messages_request, model=model, drop_params=getattr(litellm_params, "drop_params", None)
+        )
         #########################################################
         ############## BEDROCK Invoke SPECIFIC TRANSFORMATION ###
         #########################################################
@@ -822,6 +829,20 @@ class AmazonAnthropicClaudeMessagesConfig(
         if filtered_betas:
             anthropic_messages_request["anthropic_beta"] = filtered_betas
 
+        final_request: Final = OBJECT_MAPPING_ADAPTER.validate_python(anthropic_messages_request)  # pyright: ignore[reportUnknownArgumentType]  # built from the base signature's bare dict
+        gated: Final = strip_beta_gated_native_request(
+            {},
+            final_request,
+            final_request.get("messages"),
+            body_betas=filtered_betas,
+            drop_params=getattr(litellm_params, "drop_params", None),
+        )
+        log_beta_gated_strips(gated.stripped, model=model, provider="bedrock_invoke")
+        if gated.thinking_stripped:
+            anthropic_messages_request["thinking"] = gated.thinking
+        if gated.messages_stripped:
+            anthropic_messages_request["messages"] = list(gated.messages)
+
         outbound_tools: Final = tools_without_eager_input_streaming(anthropic_messages_request)
         if outbound_tools is not None:
             anthropic_messages_request["tools"] = outbound_tools

litellm/llms/bedrock/responses/transformation.py

@@ -52,6 +52,7 @@ from litellm.llms.base_llm.responses.codex_compat import drop_unsupported_tools,
 from litellm.llms.bedrock.base_aws_llm import BaseAWSLLM
 from litellm.llms.bedrock.common_utils import (
     BEDROCK_CHAT_COMPLETIONS_ROUTE_PREFIX,
+    DROP_DISABLED_REASONING_EFFORT_WARNING,
     BedrockError,
     bedrock_reasoning_effort_disabled,
     bedrock_supports_openai_responses,
@@ -173,7 +174,7 @@ def _without_disabled_reasoning_effort(
             ),
             status_code=400,
         )
-    verbose_logger.debug("Dropping unsupported `reasoning.effort=%s` for Bedrock model=%s.", effort, model)
+    verbose_logger.warning(DROP_DISABLED_REASONING_EFFORT_WARNING, "reasoning.effort", effort, model)
     rest: Final = {key: value for key, value in reasoning.items() if key != "effort"}
     without_reasoning: Final = {key: value for key, value in params.items() if key != "reasoning"}
     return {**without_reasoning, "reasoning": rest} if rest else without_reasoning

litellm/llms/cohere/embed/transformation.py

@@ -10,11 +10,13 @@ Convers
 Docs - https://docs.cohere.com/v2/reference/embed
 """
 
-from collections.abc import Iterable, Mapping, Sized
-from typing import Final, Protocol, cast
+from collections.abc import Iterable, Iterator, Mapping, Sequence, Sized
+from enum import Enum
+from itertools import chain, starmap
+from typing import Final, Protocol, TypeAlias, cast
 
 import httpx
-from pydantic import ConfigDict, TypeAdapter
+from pydantic import BaseModel, ConfigDict, Field, TypeAdapter
 
 import litellm
 from litellm import COHERE_DEFAULT_EMBEDDING_INPUT_TYPE
@@ -32,14 +34,117 @@ from litellm.utils import is_base64_encoded
 from ..common_utils import CohereError
 
 _JSON_OBJECT: Final = TypeAdapter(Mapping[str, object], config=ConfigDict(hide_input_in_errors=True))
-_EMBEDDINGS_BY_TYPE: Final = TypeAdapter(Mapping[str, Iterable[object]], config=ConfigDict(hide_input_in_errors=True))
 _DECODED_JSON: Final = TypeAdapter(object)
 
 
 class _SupportsEncode(Protocol):
     """Tokenizer handle: the embedding usage path only encodes text to measure its token length."""
 
-    def encode(self, text: str, /) -> Sized: ...
+    def encode_ordinary(self, text: str, /) -> Sized: ...
+
+
+class CohereImageTokenBilling(Enum):
+    SEPARATE_FROM_INPUT_TOKENS = "separate_from_input_tokens"
+    FOLDED_INTO_INPUT_TOKENS = "folded_into_input_tokens"
+
+
+class CohereEmbeddingRowShape(Enum):
+    OPENAI = "openai"
+    TYPED = "typed"
+
+
+class _CohereBilledUnits(BaseModel):
+    model_config = ConfigDict(frozen=True, hide_input_in_errors=True)
+
+    input_tokens: int | None = None
+    image_tokens: int | None = None
+    images: int | None = None
+
+
+class _CohereEmbedMeta(BaseModel):
+    model_config = ConfigDict(frozen=True, hide_input_in_errors=True)
+
+    billed_units: _CohereBilledUnits = Field(default_factory=_CohereBilledUnits)
+
+
+_CohereEmbeddings: TypeAlias = Mapping[str, Sequence[object]] | Sequence[object]
+
+
+class _CohereEmbedBody(BaseModel):
+    model_config = ConfigDict(frozen=True, hide_input_in_errors=True)
+
+    embeddings: _CohereEmbeddings
+    meta: _CohereEmbedMeta = Field(default_factory=_CohereEmbedMeta)
+
+
+def _embedding_rows(vectors: Sequence[object]) -> Iterator[Mapping[str, object]]:
+    return ({"object": "embedding", "index": index, "embedding": vector} for index, vector in enumerate(vectors))
+
+
+def _typed_embedding_rows(embedding_type: str, vectors: Sequence[object]) -> Iterator[Mapping[str, object]]:
+    return ({**row, "type": embedding_type} for row in _embedding_rows(vectors))
+
+
+def _response_rows(embeddings: _CohereEmbeddings, row_shape: CohereEmbeddingRowShape) -> Iterable[Mapping[str, object]]:
+    if not isinstance(embeddings, Mapping):
+        return _embedding_rows(embeddings)
+    match row_shape:
+        case CohereEmbeddingRowShape.OPENAI:
+            return chain.from_iterable(map(_embedding_rows, embeddings.values()))
+        case CohereEmbeddingRowShape.TYPED:
+            return chain.from_iterable(starmap(_typed_embedding_rows, embeddings.items()))
+
+
+def _prompt_usage(prompt_tokens: int, prompt_tokens_details: PromptTokensDetailsWrapper | None) -> Usage:
+    return Usage(
+        prompt_tokens=prompt_tokens,
+        completion_tokens=0,
+        total_tokens=prompt_tokens,
+        prompt_tokens_details=prompt_tokens_details,
+    )
+
+
+def _billed_usage(billed_units: _CohereBilledUnits, image_billing: CohereImageTokenBilling) -> Usage:
+    match image_billing:
+        case CohereImageTokenBilling.SEPARATE_FROM_INPUT_TOKENS:
+            image_units: Final = (billed_units.image_tokens or 0) + (billed_units.images or 0)
+            return _prompt_usage(
+                (billed_units.input_tokens or 0) + image_units,
+                PromptTokensDetailsWrapper(
+                    text_tokens=billed_units.input_tokens,
+                    image_tokens=image_units or None,
+                    image_count=billed_units.images,
+                ),
+            )
+        case CohereImageTokenBilling.FOLDED_INTO_INPUT_TOKENS:
+            return _prompt_usage(billed_units.input_tokens or 0, None)
+
+
+def _cohere_embed_usage(
+    billed_units: _CohereBilledUnits,
+    input_texts: Iterable[str],
+    encoding: _SupportsEncode,
+    image_billing: CohereImageTokenBilling,
+) -> Usage:
+    if billed_units.input_tokens is None and billed_units.image_tokens is None and billed_units.images is None:
+        return _prompt_usage(sum(len(encoding.encode_ordinary(text)) for text in input_texts), None)
+    return _billed_usage(billed_units, image_billing)
+
+
+def cohere_embedding_response(
+    body: Mapping[str, object],
+    model: str | None,
+    encoding: _SupportsEncode,
+    input_texts: Iterable[str],
+    image_billing: CohereImageTokenBilling,
+    row_shape: CohereEmbeddingRowShape,
+) -> EmbeddingResponse:
+    cohere_body: Final = _CohereEmbedBody.model_validate(body)
+    return EmbeddingResponse(
+        model=model,
+        data=list(_response_rows(cohere_body.embeddings, row_shape)),
+        usage=_cohere_embed_usage(cohere_body.meta.billed_units, input_texts, encoding, image_billing),
+    )
 
 
 class CohereEmbeddingConfig(BaseEmbeddingConfig):
@@ -145,44 +250,16 @@ class CohereEmbeddingConfig(BaseEmbeddingConfig):
             ),
         )
 
-    def _calculate_usage(self, input: list[str], encoding: _SupportsEncode, meta: dict) -> Usage:
-        input_tokens = 0
-
-        text_tokens: Final[int | None] = meta.get("billed_units", {}).get("input_tokens")
-
-        image_tokens: Final[int | None] = meta.get("billed_units", {}).get("images")
-
-        prompt_tokens_details: PromptTokensDetailsWrapper | None = None
-        if image_tokens is None and text_tokens is None:
-            for text in input:
-                input_tokens += len(encoding.encode(text))
-        else:
-            prompt_tokens_details = PromptTokensDetailsWrapper(
-                image_tokens=image_tokens,
-                text_tokens=text_tokens,
-            )
-            if image_tokens:
-                input_tokens += image_tokens
-            if text_tokens:
-                input_tokens += text_tokens
-
-        return Usage(
-            prompt_tokens=input_tokens,
-            completion_tokens=0,
-            total_tokens=input_tokens,
-            prompt_tokens_details=prompt_tokens_details,
-        )
-
-    def _transform_response(
+    def transform_native_embed_response(
         self,
         response: httpx.Response,
         api_key: str | None,
         logging_obj: LiteLLMLoggingObj,
-        data: dict | CohereEmbeddingRequest,
-        model_response: EmbeddingResponse,
+        data: Mapping[str, object] | CohereEmbeddingRequest,
         model: str,
         encoding: _SupportsEncode,
-        input: list,
+        input: Sequence[str],
+        row_shape: CohereEmbeddingRowShape,
     ) -> EmbeddingResponse:
         response_json: Final = _DECODED_JSON.validate_python(response.json())
         ## LOGGING
@@ -192,38 +269,15 @@ class CohereEmbeddingConfig(BaseEmbeddingConfig):
             additional_args={"complete_input_dict": data},
             original_response=response_json,
         )
-        """
-            response 
-            {
-                'object': "list",
-                'data': [
-                
-                ]
-                'model', 
-                'usage'
-            }
-        """
-        payload: Final = _JSON_OBJECT.validate_python(response_json)
-        embeddings: Final = _EMBEDDINGS_BY_TYPE.validate_python(payload["embeddings"])
-        output_data: Final = []
-        for k, embedding_list in embeddings.items():
-            for idx, embedding in enumerate(embedding_list):
-                output_data.append({"object": "embedding", "index": idx, "embedding": embedding})
-        model_response.object = "list"
-        model_response.data = output_data
-        model_response.model = model
-        input_tokens = 0
-        for text in input:
-            input_tokens += len(encoding.encode(text))
-
-        setattr(
-            model_response,
-            "usage",
-            self._calculate_usage(input, encoding, dict(_JSON_OBJECT.validate_python(payload.get("meta", {})))),
+        return cohere_embedding_response(
+            _JSON_OBJECT.validate_python(response_json),
+            model,
+            encoding,
+            input,
+            CohereImageTokenBilling.SEPARATE_FROM_INPUT_TOKENS,
+            row_shape,
         )
 
-        return model_response
-
     def transform_embedding_response(
         self,
         model: str,
@@ -235,15 +289,15 @@ class CohereEmbeddingConfig(BaseEmbeddingConfig):
         optional_params: dict,
         litellm_params: dict,
     ) -> EmbeddingResponse:
-        return self._transform_response(
+        return self.transform_native_embed_response(
             response=raw_response,
             api_key=api_key,
             logging_obj=logging_obj,
             data=request_data,
-            model_response=model_response,
             model=model,
             encoding=litellm.encoding,
             input=logging_obj.model_call_details["input"],
+            row_shape=CohereEmbeddingRowShape.OPENAI,
         )
 
     def get_error_class(self, error_message: str, status_code: int, headers: dict | httpx.Headers) -> BaseLLMException:

litellm/llms/custom_httpx/http_handler.py

@@ -764,18 +764,23 @@ class AsyncHTTPHandler:
         follow_redirects: bool | None = None,
         timeout: float | httpx.Timeout | None = None,
         max_response_bytes: int | None = None,
+        stream: bool = False,
     ):
         # Set follow_redirects to UseClientDefault if None
         _follow_redirects: Final = follow_redirects if follow_redirects is not None else USE_CLIENT_DEFAULT
+        query: Final = httpx.QueryParams(params).merge(HTTPHandler.extract_query_params(url))
+        request_headers: Final = httpx.Headers(headers)
 
-        params = params or {}
-        params.update(HTTPHandler.extract_query_params(url))
+        if stream:
+            return await self._send_streaming_get(
+                url, params=query, headers=request_headers, follow_redirects=follow_redirects, timeout=timeout
+            )
 
         if max_response_bytes is not None:
             return await self._get_with_response_limit(
                 url,
-                params=httpx.QueryParams(params),
-                headers=httpx.Headers(headers),
+                params=query,
+                headers=request_headers,
                 max_bytes=max_response_bytes,
                 follow_redirects=self.client.follow_redirects if follow_redirects is None else follow_redirects,
                 timeout=self.client.timeout if timeout is None else httpx.Timeout(timeout),
@@ -783,13 +788,41 @@ class AsyncHTTPHandler:
 
         response: Final = await self.client.get(
             url,
-            params=params,
-            headers=headers,
+            params=query,
+            headers=request_headers,
             follow_redirects=_follow_redirects,
             timeout=timeout if timeout is not None else USE_CLIENT_DEFAULT,
         )
         return response
 
+    async def _send_streaming_get(
+        self,
+        url: str,
+        *,
+        params: httpx.QueryParams,
+        headers: httpx.Headers,
+        follow_redirects: bool | None,
+        timeout: float | httpx.Timeout | None,
+    ) -> httpx.Response:
+        start_time: Final = time.time()
+        request: Final = self.client.build_request(
+            "GET", url, params=params, headers=headers, timeout=USE_CLIENT_DEFAULT if timeout is None else timeout
+        )
+        try:
+            response: Final = await self.client.send(
+                request,
+                stream=True,
+                follow_redirects=USE_CLIENT_DEFAULT if follow_redirects is None else follow_redirects,
+            )
+            _anchor_handler_to(response, self)
+            response.raise_for_status()
+        except httpx.TimeoutException as e:
+            _raise_timeout(e, start_time)
+        except httpx.HTTPStatusError as e:
+            await _raise_masked_async_error(e, True)
+        else:
+            return response
+
     async def _get_with_response_limit(
         self,
         url: str,
@@ -1468,15 +1501,20 @@ class HTTPHandler:
         headers: Mapping[str, Any] | None = None,
         follow_redirects: bool | None = None,
         timeout: float | httpx.Timeout | None = None,
+        stream: bool = False,
     ):
         # Set follow_redirects to UseClientDefault if None
         _follow_redirects: Final = follow_redirects if follow_redirects is not None else USE_CLIENT_DEFAULT
-        params = params or {}
-        params.update(self.extract_query_params(url))
+        query: Final = httpx.QueryParams(params).merge(self.extract_query_params(url))
+
+        if stream:
+            return self._send_streaming_get(
+                url, params=query, headers=headers, follow_redirects=follow_redirects, timeout=timeout
+            )
 
         response: Final = self.client.get(
             url,
-            params=params,
+            params=query,
             headers=headers,
             follow_redirects=_follow_redirects,
             timeout=timeout if timeout is not None else USE_CLIENT_DEFAULT,
@@ -1484,6 +1522,34 @@ class HTTPHandler:
 
         return response
 
+    def _send_streaming_get(
+        self,
+        url: str,
+        *,
+        params: httpx.QueryParams,
+        headers: Mapping[str, str] | None,
+        follow_redirects: bool | None,
+        timeout: float | httpx.Timeout | None,
+    ) -> httpx.Response:
+        start_time: Final = time.time()
+        request: Final = self.client.build_request(
+            "GET", url, params=params, headers=headers, timeout=USE_CLIENT_DEFAULT if timeout is None else timeout
+        )
+        try:
+            response: Final = self.client.send(
+                request,
+                stream=True,
+                follow_redirects=USE_CLIENT_DEFAULT if follow_redirects is None else follow_redirects,
+            )
+            _anchor_handler_to(response, self)
+            response.raise_for_status()
+        except httpx.TimeoutException as e:
+            _raise_timeout(e, start_time)
+        except httpx.HTTPStatusError as e:
+            _raise_masked_sync_error(e, True)
+        else:
+            return response
+
     @staticmethod
     def extract_query_params(url: str) -> dict[str, str]:
         """

litellm/llms/custom_httpx/llm_http_handler.py

@@ -50,7 +50,11 @@ from litellm.litellm_core_utils.audio_utils.subtitle_utils import (
     SUBTITLE_RESPONSE_FORMATS,
     synthesize_subtitle_document,
 )
-from litellm.litellm_core_utils.core_helpers import as_str_mapping, set_provider_response_headers_in_hidden_params
+from litellm.litellm_core_utils.core_helpers import (
+    as_str_mapping,
+    qualify_agentic_followup_model,
+    set_provider_response_headers_in_hidden_params,
+)
 from litellm.litellm_core_utils.dot_notation_indexing import delete_nested_value
 from litellm.litellm_core_utils.get_litellm_params import AWS_CREDENTIAL_KWARGS_KEYS
 from litellm.litellm_core_utils.get_llm_provider_logic import get_provider_affinity_header
@@ -386,6 +390,18 @@ def _signed_body_kwargs(*, signed_body: bytes | None, data: dict[str, object]) -
     return {"json": data}
 
 
+def _answered_json(response: httpx.Response) -> bool:
+    return "application/json" in response.headers.get("content-type", "").lower()
+
+
+def _stream_query_params(stream: bool, starting_after: int | None) -> Mapping[str, object]:
+    if not stream:
+        return {}
+    if starting_after is None:
+        return {"stream": "true"}
+    return {"stream": "true", "starting_after": starting_after}
+
+
 def _google_genai_streaming_hidden_params(
     *,
     api_base: str,
@@ -2565,6 +2581,7 @@ class BaseLLMHTTPHandler:
                 "preset_cache_key": None,
                 "stream_response": {},
                 "model_info": kwargs.get("model_info"),
+                **litellm_params.per_second_pricing(),
                 **vertex_location_params,
                 **anthropic_messages_optional_request_params,
             },
@@ -3524,12 +3541,20 @@ class BaseLLMHTTPHandler:
         logging_obj: LiteLLMLoggingObj,
         custom_llm_provider: str | None = None,
         extra_headers: dict[str, object] | None = None,
+        extra_query: Mapping[str, object] | None = None,
         extra_body: dict[str, object] | None = None,
+        stream: bool = False,
+        starting_after: int | None = None,
+        litellm_metadata: Mapping[str, object] = MappingProxyType({}),
         timeout: float | httpx.Timeout | None = None,
         client: HTTPHandler | AsyncHTTPHandler | None = None,
         _is_async: bool = False,
         shared_session: Optional["ClientSession"] = None,
-    ) -> ResponsesAPIResponse | Coroutine[object, object, ResponsesAPIResponse]:
+    ) -> (
+        ResponsesAPIResponse
+        | SyncResponsesAPIStreamingIterator
+        | Coroutine[object, object, ResponsesAPIResponse | ResponsesAPIStreamingIterator]
+    ):
         """
         Get a response by ID
         Uses GET /v1/responses/{response_id} endpoint in the responses API
@@ -3542,7 +3567,11 @@ class BaseLLMHTTPHandler:
                 logging_obj=logging_obj,
                 custom_llm_provider=custom_llm_provider,
                 extra_headers=extra_headers,
+                extra_query=extra_query,
                 extra_body=extra_body,
+                stream=stream,
+                starting_after=starting_after,
+                litellm_metadata=litellm_metadata,
                 timeout=timeout,
                 client=client,
                 shared_session=shared_session,
@@ -3571,20 +3600,23 @@ class BaseLLMHTTPHandler:
             litellm_params=litellm_params,
             headers=headers,
         )
+        query: Final = {**data, **(extra_query or {}), **_stream_query_params(stream, starting_after)}
 
         ## LOGGING
         logging_obj.pre_call(
             input="",
             api_key="",
             additional_args={
-                "complete_input_dict": data,
+                "complete_input_dict": query,
                 "api_base": api_base,
                 "headers": headers,
             },
         )
 
         try:
-            response: Final = sync_httpx_client.get(url=url, headers=headers, params=data)
+            response: Final = sync_httpx_client.get(
+                url=url, headers=headers, params=query, stream=stream, timeout=timeout if stream else None
+            )
             response.raise_for_status()
         except Exception as e:
             raise self._handle_error(
@@ -3592,6 +3624,20 @@ class BaseLLMHTTPHandler:
                 provider_config=responses_api_provider_config,
             )
 
+        if stream and not _answered_json(response):
+            return SyncResponsesAPIStreamingIterator(
+                response=response,
+                model="",
+                logging_obj=logging_obj,
+                responses_api_provider_config=responses_api_provider_config,
+                litellm_metadata=dict(litellm_metadata),
+                custom_llm_provider=custom_llm_provider,
+                request_data={"litellm_params": dict(litellm_params)},
+                replays_stored_stream=True,
+            )
+
+        if stream:
+            response.read()
         return responses_api_provider_config.transform_get_response_api_response(
             raw_response=response,
             logging_obj=logging_obj,
@@ -3605,11 +3651,15 @@ class BaseLLMHTTPHandler:
         logging_obj: LiteLLMLoggingObj,
         custom_llm_provider: str | None = None,
         extra_headers: dict[str, object] | None = None,
+        extra_query: Mapping[str, object] | None = None,
         extra_body: dict[str, object] | None = None,
+        stream: bool = False,
+        starting_after: int | None = None,
+        litellm_metadata: Mapping[str, object] = MappingProxyType({}),
         timeout: float | httpx.Timeout | None = None,
         client: HTTPHandler | AsyncHTTPHandler | None = None,
         shared_session: Optional["ClientSession"] = None,
-    ) -> ResponsesAPIResponse:
+    ) -> ResponsesAPIResponse | ResponsesAPIStreamingIterator:
         """
         Async version of get_responses
         """
@@ -3647,20 +3697,23 @@ class BaseLLMHTTPHandler:
             litellm_params=litellm_params,
             headers=headers,
         )
+        query: Final = {**data, **(extra_query or {}), **_stream_query_params(stream, starting_after)}
 
         ## LOGGING
         logging_obj.pre_call(
             input="",
             api_key="",
             additional_args={
-                "complete_input_dict": data,
+                "complete_input_dict": query,
                 "api_base": api_base,
                 "headers": headers,
             },
         )
 
         try:
-            response: Final = await async_httpx_client.get(url=url, headers=headers, params=data)
+            response: Final = await async_httpx_client.get(
+                url=url, headers=headers, params=query, stream=stream, timeout=timeout if stream else None
+            )
             response.raise_for_status()
         except Exception as e:
             verbose_logger.debug("Error retrieving response: %s", e)
@@ -3669,6 +3722,20 @@ class BaseLLMHTTPHandler:
                 provider_config=responses_api_provider_config,
             )
 
+        if stream and not _answered_json(response):
+            return ResponsesAPIStreamingIterator(
+                response=response,
+                model="",
+                logging_obj=logging_obj,
+                responses_api_provider_config=responses_api_provider_config,
+                litellm_metadata=dict(litellm_metadata),
+                custom_llm_provider=custom_llm_provider,
+                request_data={"litellm_params": dict(litellm_params)},
+                replays_stored_stream=True,
+            )
+
+        if stream:
+            await response.aread()
         return responses_api_provider_config.transform_get_response_api_response(
             raw_response=response,
             logging_obj=logging_obj,
@@ -6213,9 +6280,7 @@ class BaseLLMHTTPHandler:
         if patch.messages is None:
             raise ValueError("Agentic loop plan missing patched messages")
 
-        full_model_name = patch.model or model
-        if "/" not in full_model_name:
-            full_model_name = f"{custom_llm_provider}/{full_model_name}"
+        full_model_name: Final = qualify_agentic_followup_model(patch.model, model, custom_llm_provider)
 
         optional_params_for_followup: Final = dict(optional_params)
         optional_params_for_followup.update(patch.optional_params)

litellm/llms/databricks/chat/transformation.py

@@ -10,6 +10,7 @@ from typing import TYPE_CHECKING, Any, Final, Literal, cast, overload
 import httpx
 from pydantic import BaseModel, ConfigDict, TypeAdapter
 
+from litellm.anthropic_beta_gated_fields import log_beta_gated_strips, strip_beta_gated_thinking
 from litellm.constants import RESPONSE_FORMAT_TOOL_NAME
 from litellm.litellm_core_utils.llm_response_utils.convert_dict_to_response import (
     handle_invalid_parallel_tool_calls,
@@ -514,6 +515,14 @@ class DatabricksConfig(DatabricksBase, OpenAILikeChatConfig, AnthropicConfig):
         self.update_optional_params_with_thinking_tokens(
             non_default_params=non_default_params, optional_params=mapped_params
         )
+        gated_thinking: Final = strip_beta_gated_thinking(
+            mapped_params.get("thinking"),  # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType]  # mapped_params is the base config's bare dict
+            frozenset(),
+            drop_params=drop_params,
+        )
+        if gated_thinking.stripped:
+            log_beta_gated_strips(gated_thinking.stripped, model=model, provider="databricks")
+            mapped_params["thinking"] = gated_thinking.thinking
 
         return mapped_params
 

litellm/llms/databricks/messages/transformation.py

@@ -144,7 +144,7 @@ class DatabricksAnthropicMessagesConfig(DatabricksBase, AnthropicMessagesConfig)
             ("content-type" if key.lower() == "content-type" else key): value
             for key, value in {**auth_headers, **client_anthropic_headers}.items()
         }
-        finalized_headers: Final[dict[str, str]] = self._finalize_messages_headers(  # pyright: ignore[reportUnknownMemberType, reportUnknownVariableType]  # the shared helper returns a bare dict  # mutable-ok: becomes the outbound headers
+        finalized_headers: Final[dict[str, str]] = self.finalize_messages_headers(  # pyright: ignore[reportUnknownMemberType, reportUnknownVariableType]  # the shared helper returns a bare dict  # mutable-ok: becomes the outbound headers
             headers=normalized_headers,
             optional_params=optional_params,
             messages=messages,
@@ -191,6 +191,8 @@ class DatabricksAnthropicMessagesConfig(DatabricksBase, AnthropicMessagesConfig)
             litellm_params=litellm_params,
             headers=headers,
         )
-        self._normalize_system_role_messages(anthropic_messages_request, model=model)  # pyright: ignore[reportUnknownMemberType]  # the shared helper takes a bare dict
+        self._normalize_system_role_messages(  # pyright: ignore[reportUnknownMemberType]  # the shared helper takes a bare dict
+            anthropic_messages_request, model=model, drop_params=getattr(litellm_params, "drop_params", None)
+        )
         self._remove_scope_from_cache_control(anthropic_messages_request)  # pyright: ignore[reportUnknownMemberType]  # the shared helper takes a bare dict
         return anthropic_messages_request

litellm/llms/fireworks_ai/responses/transformation.py

@@ -1,10 +1,9 @@
-from collections.abc import Mapping, Sequence
+from collections.abc import Mapping
 from types import MappingProxyType
 from typing import TYPE_CHECKING, Final
 from urllib.parse import unquote
 
 import httpx
-from openai.types.responses import EasyInputMessageParam, ResponseInputContentParam, ResponseInputItemParam
 
 from litellm.litellm_core_utils.url_utils import join_api_base
 from litellm.llms.fireworks_ai.common_utils import (
@@ -15,12 +14,18 @@ from litellm.llms.fireworks_ai.common_utils import (
     with_fireworks_session_affinity,
     without_caller_user,
 )
+from litellm.llms.openai.responses.instruction_items import developer_items_as_system, instruction_items_folded
 from litellm.llms.openai.responses.transformation import OpenAIResponsesAPIConfig
 from litellm.secret_managers.main import get_secret_str
 from litellm.types.llms.openai import ResponseInputParam
 from litellm.types.responses.main import DeleteResponseResult
 from litellm.types.router import GenericLiteLLMParams
 from litellm.types.utils import LlmProviders
+from litellm.utils import (
+    declared_mid_conversation_system_support,
+    deployment_declared_mid_conversation_system_support,
+    deployment_model_info,
+)
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
@@ -45,90 +50,6 @@ def _forwarded_user_params(litellm_params: GenericLiteLLMParams) -> Mapping[str,
     )
 
 
-_INSTRUCTION_ROLES: Final = frozenset({"system", "developer"})
-
-
-def _role(item: ResponseInputItemParam) -> str | None:
-    match item:
-        case {"role": str(role)}:
-            return role
-        case _:
-            return None
-
-
-def _developer_item_as_system(item: ResponseInputItemParam) -> ResponseInputItemParam:
-    if "role" not in item or item["role"] != "developer":
-        return item
-    return EasyInputMessageParam(role="system", content=item["content"], type="message")
-
-
-def _developer_items_as_system(input: str | ResponseInputParam) -> str | ResponseInputParam:
-    if isinstance(input, str):
-        return input
-    return [_developer_item_as_system(item) for item in input]
-
-
-def _text_part(part: ResponseInputContentParam) -> str | None:
-    match part:
-        case {"type": "input_text", "text": str(text)}:
-            return text
-        case _:
-            return None
-
-
-def _text_only_content(item: ResponseInputItemParam) -> str | None:
-    match item:
-        case {"role": "system" | "developer", "content": str(text)}:
-            return text
-        case {"role": "system" | "developer", "content": [*parts]}:
-            texts: Final = tuple(map(_text_part, parts))
-            return None if any(text is None for text in texts) else "\n\n".join(text for text in texts if text)
-        case _:
-            return None
-
-
-def _leading_instruction_block_length(roles: Sequence[str | None]) -> int:
-    return next((index for index, role in enumerate(roles) if role not in _INSTRUCTION_ROLES), len(roles))
-
-
-def _closing_instruction_block_start(roles: Sequence[str | None], leading_length: int) -> int:
-    last_conversation_index: Final = next(
-        (index for index in range(len(roles) - 1, leading_length - 1, -1) if roles[index] not in _INSTRUCTION_ROLES),
-        None,
-    )
-    if last_conversation_index is None or roles[last_conversation_index] != "assistant":
-        return len(roles)
-    return last_conversation_index + 1
-
-
-def _hoisted_indices(roles: Sequence[str | None]) -> tuple[int, ...]:
-    leading_length: Final = _leading_instruction_block_length(roles)
-    closing_start: Final = _closing_instruction_block_start(roles, leading_length)
-    return tuple(
-        index for index, role in enumerate(roles[:closing_start]) if index < leading_length or role == "developer"
-    )
-
-
-def _with_instruction_items_folded(
-    input: str | ResponseInputParam, instructions: str | None
-) -> tuple[str | None, str | ResponseInputParam]:
-    if isinstance(input, str):
-        return instructions, input
-    items: Final = tuple(input)
-    folded: Final = MappingProxyType(
-        {
-            index: text
-            for index in _hoisted_indices(tuple(map(_role, items)))
-            if (text := _text_only_content(items[index])) is not None
-        }
-    )
-    joined: Final = "\n\n".join(chunk for chunk in (instructions, *folded.values()) if chunk)
-    return (
-        instructions if not folded else joined or None,
-        [_developer_item_as_system(item) for index, item in enumerate(items) if index not in folded],
-    )
-
-
 class FireworksAIResponsesAPIConfig(OpenAIResponsesAPIConfig):
     @property
     def custom_llm_provider(self) -> LlmProviders:
@@ -158,29 +79,28 @@ class FireworksAIResponsesAPIConfig(OpenAIResponsesAPIConfig):
         self,
         model: str,
         input: str | ResponseInputParam,
-        response_api_optional_request_params: dict,  # mutable-ok: overrides the base class signature
+        response_api_optional_request_params: dict[str, object],  # mutable-ok: overrides the base class signature
         litellm_params: GenericLiteLLMParams,
-        headers: dict,  # mutable-ok: overrides the base class signature
-    ) -> dict:  # mutable-ok: overrides the base class signature
-        instructions_param: Final[object] = response_api_optional_request_params.get("instructions")
+        headers: dict[str, str],  # mutable-ok: overrides the base class signature
+    ) -> dict[str, object]:  # mutable-ok: overrides the base class signature
         validated_input: Final = self._validate_input_param(input)
-        instructions, folded_input = (
-            _with_instruction_items_folded(validated_input, instructions_param)
-            if isinstance(instructions_param, str | None)
-            else (instructions_param, _developer_items_as_system(validated_input))
-        )
-        instruction_entries: Final = () if instructions is None else (("instructions", instructions),)
-        folded_params: Final = {
-            key: value
-            for key, value in (
-                *((key, value) for key, value in response_api_optional_request_params.items() if key != "instructions"),
-                *instruction_entries,
+        model_info: Final = deployment_model_info(litellm_params)
+        declared: Final = declared_mid_conversation_system_support(model, self.custom_llm_provider.value, model_info)
+        folded: Final = (
+            None
+            if deployment_declared_mid_conversation_system_support(model_info) is True
+            else instruction_items_folded(
+                validated_input, response_api_optional_request_params, "user" if declared is False else "system"
             )
-        }
+        )
+        placed_input, placed_params = folded or (
+            developer_items_as_system(validated_input),
+            response_api_optional_request_params,
+        )
         request: Final = super().transform_responses_api_request(
             model=resolve_fireworks_resource_name(model),
-            input=folded_input,
-            response_api_optional_request_params=folded_params,
+            input=placed_input,
+            response_api_optional_request_params=dict(placed_params),
             litellm_params=litellm_params,
             headers=headers,
         )

litellm/llms/github_copilot/authenticator.py

@@ -1,11 +1,12 @@
 import json
 import os
 import time
+from dataclasses import dataclass
 from datetime import datetime
-from typing import Any, Final
+from typing import Any, Final, TypeVar
 
 import httpx
-from pydantic import ConfigDict, TypeAdapter, with_config
+from pydantic import BaseModel, ConfigDict, TypeAdapter, with_config
 from typing_extensions import NotRequired, ReadOnly, TypedDict
 
 from litellm._logging import verbose_logger
@@ -13,7 +14,6 @@ from litellm.litellm_core_utils.asyncify import can_block_current_thread
 from litellm.llms.custom_httpx.http_handler import get_httpx_client
 
 from .common_utils import (
-    APIKeyExpiredError,
     GetAccessTokenError,
     GetAPIKeyError,
     GetDeviceCodeError,
@@ -27,6 +27,44 @@ DEFAULT_GITHUB_ACCESS_TOKEN_URL: Final = "https://github.com/login/oauth/access_
 DEFAULT_GITHUB_API_KEY_URL: Final = "https://api.github.com/copilot_internal/v2/token"
 
 
+class CopilotEndpoints(BaseModel):
+    model_config = ConfigDict(frozen=True, strict=True, extra="ignore", hide_input_in_errors=True)
+
+    api: str | None = None
+
+
+class CopilotCachedKey(BaseModel):
+    model_config = ConfigDict(frozen=True, strict=True, extra="ignore", hide_input_in_errors=True)
+
+    token: str
+    expires_at: float
+    endpoints: CopilotEndpoints | None = None
+
+
+class CopilotCachedEndpoints(BaseModel):
+    model_config = ConfigDict(frozen=True, strict=True, extra="ignore", hide_input_in_errors=True)
+
+    endpoints: CopilotEndpoints | None = None
+
+
+ApiKeyFileView = TypeVar("ApiKeyFileView", CopilotCachedKey, CopilotCachedEndpoints)
+
+
+@dataclass(frozen=True, slots=True)
+class _UnsavedApiKey:
+    access_token: str
+    key: CopilotCachedKey
+
+
+_UNSAVED_API_KEYS: Final[dict[str, _UnsavedApiKey]] = {}  # mutable-ok: refreshed keys api-key.json could not hold
+
+
+def _read_token_file(path: str) -> str:
+    with open(path, "rb") as f:
+        raw: Final = f.read()
+    return raw.decode(json.detect_encoding(raw))
+
+
 @with_config(ConfigDict(extra="allow", strict=True))
 class _DeviceCode(TypedDict):
     device_code: ReadOnly[str]
@@ -87,12 +125,13 @@ class Authenticator:
             GetAccessTokenError: If unable to obtain an access token after retries.
         """
         try:
-            with open(self.access_token_file, "r") as f:
-                access_token = f.read().strip()
-                if access_token:
-                    return access_token
+            saved_token: Final = _read_token_file(self.access_token_file).strip()
+            if saved_token:
+                return saved_token
         except OSError:
             verbose_logger.warning("No existing access token found or error reading file")
+        except ValueError as e:
+            verbose_logger.warning("Ignoring %s, not a text access token: %s", self.access_token_file, e)
 
         if not can_block_current_thread():
             raise GetAccessTokenError(
@@ -135,64 +174,83 @@ class Authenticator:
         Raises:
             GetAPIKeyError: If unable to obtain an API key.
         """
-        try:
-            with open(self.api_key_file, "r") as f:
-                api_key_info = json.load(f)
-                if api_key_info.get("expires_at", 0) > datetime.now().timestamp():
-                    return api_key_info.get("token")
-                else:
-                    verbose_logger.warning("API key expired, refreshing")
-                    raise APIKeyExpiredError(
-                        message="API key expired",
-                        status_code=401,
-                    )
-        except OSError:
-            verbose_logger.warning("No API key file found or error opening file")
-        except (json.JSONDecodeError, KeyError) as e:
-            verbose_logger.warning("Error reading API key from file: %s", e)
-        except APIKeyExpiredError:
-            pass  # Already logged in the try block
+        unsaved: Final = self._unsaved_api_key()
+        if unsaved is not None:
+            return unsaved.token
+        cached: Final = self._read_api_key_file(CopilotCachedKey)
+        if cached is not None and cached.expires_at > datetime.now().timestamp():
+            return cached.token
+        if cached is not None:
+            verbose_logger.warning("API key expired, refreshing")
 
         try:
-            api_key_info = self._refresh_api_key()
-            with open(self.api_key_file, "w") as f:
-                json.dump(api_key_info, f)
-            token: Final = api_key_info.get("token")
-            if token:
-                return token
-            else:
-                raise GetAPIKeyError(
-                    message="API key response missing token",
-                    status_code=401,
-                )
-        except OSError as e:
-            verbose_logger.error("Error saving API key to file: %s", e)
-            raise GetAPIKeyError(
-                message=f"Failed to save API key: {e}",
-                status_code=500,
-            )
+            api_key_info: Final = self._refresh_api_key()
         except RefreshAPIKeyError as e:
             raise GetAPIKeyError(
                 message=f"Failed to refresh API key: {e}",
                 status_code=401,
+            ) from e
+        serialized: Final = json.dumps(api_key_info)
+        try:
+            with open(self.api_key_file, "w") as f:
+                f.write(serialized)
+        except OSError as e:
+            verbose_logger.warning(
+                "Keeping the refreshed API key in memory, it could not be saved to %s: %s", self.api_key_file, e
             )
+            self._keep_unsaved_api_key(serialized)
+        token: Final = api_key_info.get("token")
+        if not token:
+            raise GetAPIKeyError(
+                message="API key response missing token",
+                status_code=401,
+            )
+        return token
 
     def get_api_base(self) -> str | None:
         """
-        Get the API endpoint from the api-key.json file.
+        Get the API endpoint of the current API key.
 
         Returns:
             Optional[str]: The GitHub Copilot API endpoint, or None if not found.
         """
-        try:
-            with open(self.api_key_file, "r") as f:
-                api_key_info: Final = json.load(f)
-                endpoints: Final = api_key_info.get("endpoints", {})
-                api_endpoint: Final = endpoints.get("api")
-                return api_endpoint
-        except (OSError, json.JSONDecodeError, KeyError) as e:
-            verbose_logger.warning("Error reading API endpoint from file: %s", e)
+        cached: Final = self._unsaved_api_key() or self._read_api_key_file(CopilotCachedEndpoints)
+        if cached is None or cached.endpoints is None:
             return None
+        return cached.endpoints.api
+
+    def _read_api_key_file(self, view: type[ApiKeyFileView]) -> ApiKeyFileView | None:
+        try:
+            return view.model_validate_json(_read_token_file(self.api_key_file))
+        except OSError:
+            verbose_logger.warning("No API key file found or error opening file")
+            return None
+        except ValueError as e:
+            verbose_logger.warning("Ignoring %s, not a Copilot API key file: %s", self.api_key_file, e)
+            return None
+
+    def _stored_access_token(self) -> str | None:
+        try:
+            access_token: Final = _read_token_file(self.access_token_file).strip()
+        except (OSError, ValueError):
+            return None
+        return access_token or None
+
+    def _unsaved_api_key(self) -> CopilotCachedKey | None:
+        unsaved: Final = _UNSAVED_API_KEYS.get(self.api_key_file)
+        if unsaved is None or unsaved.key.expires_at <= time.time():
+            return None
+        return unsaved.key if unsaved.access_token == self._stored_access_token() else None
+
+    def _keep_unsaved_api_key(self, serialized: str) -> None:
+        access_token: Final = self._stored_access_token()
+        if access_token is None:
+            return
+        try:
+            key: Final = CopilotCachedKey.model_validate_json(serialized)
+        except ValueError:
+            return
+        _UNSAVED_API_KEYS[self.api_key_file] = _UnsavedApiKey(access_token=access_token, key=key)
 
     def _refresh_api_key(self) -> dict[str, Any]:
         """

litellm/llms/hosted_vllm/responses/transformation.py

@@ -9,10 +9,13 @@ the chat completions → responses conversion pipeline.
 from typing import Final
 
 from litellm.litellm_core_utils.url_utils import join_api_base
+from litellm.llms.openai.responses.instruction_items import instruction_items_folded
 from litellm.llms.openai.responses.transformation import OpenAIResponsesAPIConfig
 from litellm.secret_managers.main import get_secret_str
+from litellm.types.llms.openai import ResponseInputParam
 from litellm.types.router import GenericLiteLLMParams
 from litellm.types.utils import LlmProviders
+from litellm.utils import declared_mid_conversation_system_support, deployment_model_info
 
 
 class HostedVLLMResponsesAPIConfig(OpenAIResponsesAPIConfig):
@@ -69,6 +72,31 @@ class HostedVLLMResponsesAPIConfig(OpenAIResponsesAPIConfig):
 
         return join_api_base(api_base, "v1/responses")
 
+    def transform_responses_api_request(
+        self,
+        model: str,
+        input: str | ResponseInputParam,
+        response_api_optional_request_params: dict[str, object],  # mutable-ok: overrides the base class signature
+        litellm_params: GenericLiteLLMParams,
+        headers: dict[str, str],  # mutable-ok: overrides the base class signature
+    ) -> dict[str, object]:  # mutable-ok: overrides the base class signature
+        declared: Final = declared_mid_conversation_system_support(
+            model, self.custom_llm_provider.value, deployment_model_info(litellm_params)
+        )
+        folded: Final = (
+            instruction_items_folded(self._validate_input_param(input), response_api_optional_request_params, "user")
+            if declared is False
+            else None
+        )
+        placed_input, placed_params = folded or (input, response_api_optional_request_params)
+        return super().transform_responses_api_request(
+            model=model,
+            input=placed_input,
+            response_api_optional_request_params=dict(placed_params),
+            litellm_params=litellm_params,
+            headers=headers,
+        )
+
     def supports_native_websocket(self) -> bool:
         """Hosted vLLM does not support native WebSocket for Responses API"""
         return False

litellm/llms/manus/responses/transformation.py

@@ -241,7 +241,7 @@ class ManusResponsesAPIConfig(OpenAIResponsesAPIConfig):
         api_base: str,
         litellm_params: GenericLiteLLMParams,
         headers: dict,
-    ) -> tuple[str, dict]:
+    ) -> tuple[str, dict[str, str]]:
         """
         Transform the get response API request into a URL and data.
 
@@ -252,7 +252,7 @@ class ManusResponsesAPIConfig(OpenAIResponsesAPIConfig):
         """
         encoded_response_id: Final = encode_url_path_segment(response_id, field_name="response_id")
         url: Final = f"{api_base}/{encoded_response_id}"
-        data: Final[dict] = {}
+        data: Final[dict[str, str]] = {}
         return url, data
 
     def transform_get_response_api_response(

litellm/llms/openai/chat/gpt_reasoning_transformation.py

@@ -5,8 +5,10 @@ from collections.abc import Collection
 from typing import ClassVar, Final
 
 import litellm
+from litellm._logging import verbose_logger
 from litellm.utils import (
     declared_value_factory,
+    has_catalog_entry,
     is_explicitly_disabled_factory,
     supports_factory,
 )
@@ -16,6 +18,25 @@ from .gpt_transformation import OpenAIGPTConfig
 _GPT_SERIES_VERSION: Final = re.compile(r"^gpt-(\d+)(?:\.(\d+))?(?=[.-]|$)")
 
 
+def refused_reasoning_effort_warning(*, suggest_declared_levels: bool, suggest_base_model: bool) -> str:
+    return "".join(
+        (
+            "Dropping %s=%r for model=%r (drop_params=True) because the model map does not mark that level "
+            "supported for %r.",
+            (
+                " If the deployment supports it, list every level it supports in model_info.reasoning_effort_levels."
+                if suggest_declared_levels
+                else ""
+            ),
+            (
+                " If the deployment name is not its underlying model, set model_info.base_model to that model."
+                if suggest_base_model
+                else ""
+            ),
+        )
+    )
+
+
 def _catalogue_declares_default_effort() -> bool:
     """Whether the loaded cost map carries default_reasoning_effort for ANY entry.
 
@@ -93,6 +114,7 @@ class OpenAIGPT5Config(OpenAIGPTConfig):
 
     OPT_IN_REASONING_EFFORTS: ClassVar[frozenset[str]] = frozenset(("xhigh",))
     ROUTE_SPECIFIC_REASONING_EFFORTS: ClassVar[frozenset[str]] = frozenset(("max",))
+    LOOKS_UP_BASE_MODEL: ClassVar[bool] = False
 
     @classmethod
     def is_model_gpt_5_model(cls, model: str) -> bool:
@@ -307,7 +329,7 @@ class OpenAIGPT5Config(OpenAIGPTConfig):
         model: str,
         drop_params: bool,
         *,
-        declared_reasoning_efforts: Collection[str] = (),
+        declared_reasoning_efforts: Collection[str] | None = None,
     ) -> dict:
         if self.is_model_gpt_5_search_model(model):
             if "max_tokens" in non_default_params:
@@ -338,10 +360,21 @@ class OpenAIGPT5Config(OpenAIGPTConfig):
 
         if (
             requested_effort is not None
-            and requested_effort not in declared_reasoning_efforts
+            and requested_effort not in (declared_reasoning_efforts or ())
             and self.refuses_reasoning_effort_level(model, requested_effort)
         ):
             if litellm.drop_params or drop_params:
+                verbose_logger.warning(
+                    refused_reasoning_effort_warning(
+                        suggest_declared_levels=declared_reasoning_efforts is not None,
+                        suggest_base_model=self.LOOKS_UP_BASE_MODEL
+                        and not has_catalog_entry(self._model_map_lookup_name(model), None),
+                    ),
+                    "reasoning_effort",
+                    requested_effort,
+                    model,
+                    self._model_map_lookup_name(model),
+                )
                 non_default_params.pop("reasoning_effort", None)
                 optional_params.pop("reasoning_effort", None)
             else:

litellm/llms/openai/chat/gpt_transformation.py

@@ -36,7 +36,11 @@ from litellm.litellm_core_utils.prompt_templates.image_handling import (
 )
 from litellm.litellm_core_utils.url_utils import join_api_base
 from litellm.llms.base_llm.base_model_iterator import BaseModelResponseIterator
-from litellm.llms.base_llm.base_utils import BaseLLMModelInfo
+from litellm.llms.base_llm.base_utils import (
+    BaseLLMModelInfo,
+    hoist_developer_messages_into_leading_system_message,
+    map_developer_role_to_system_role,
+)
 from litellm.llms.base_llm.chat.transformation import BaseConfig, BaseLLMException
 from litellm.secret_managers.main import get_secret_str
 from litellm.types.llms.openai import (
@@ -58,7 +62,7 @@ from litellm.types.utils import (
     ModelResponseStream,
     chat_completion_tool_call_from_dict,
 )
-from litellm.utils import convert_to_model_response_object
+from litellm.utils import convert_to_model_response_object, declared_mid_conversation_system_support
 
 from ..common_utils import OpenAIError
 from ..workload_identity import get_workload_identity_bearer_token, resolve_openai_workload_identity_config
@@ -75,6 +79,45 @@ else:
 
 
 _NO_TOOLS_UPDATE: Final[Mapping[str, object]] = MappingProxyType({})
+
+
+def targets_openai_hosted_endpoint(custom_llm_provider: str | None, api_base: str | None) -> bool:
+    if custom_llm_provider != "openai":
+        return False
+    resolved_api_base: Final = (
+        api_base or litellm.api_base or os.getenv("OPENAI_BASE_URL") or os.getenv("OPENAI_API_BASE")
+    )
+    if not resolved_api_base:
+        return True
+    hostname: Final = urlparse(resolved_api_base).hostname
+    if hostname is None:
+        return True
+    return hostname == "openai.com" or hostname.endswith(".openai.com")
+
+
+def _roles(messages: Sequence[AllMessageValues]) -> tuple[str, ...]:
+    return tuple(message["role"] for message in messages)
+
+
+def translate_developer_role_for_openai_compatible_endpoint(
+    messages: Sequence[AllMessageValues],
+    *,
+    model: str,
+    custom_llm_provider: str | None,
+    api_base: str | None,
+    model_info: object,
+) -> Sequence[AllMessageValues]:
+    in_place: Final = map_developer_role_to_system_role(messages=messages)
+    if targets_openai_hosted_endpoint(custom_llm_provider, api_base):
+        return in_place
+    placed: Final = hoist_developer_messages_into_leading_system_message(messages=messages)
+    if _roles(placed) == _roles(in_place):
+        return in_place
+    if declared_mid_conversation_system_support(model, custom_llm_provider, model_info) is not False:
+        return in_place
+    return placed
+
+
 _OBJECT_MAPPING: Final[TypeAdapter[Mapping[str, object]]] = TypeAdapter(Mapping[str, object])
 _CHOICE_MAPPINGS: Final = TypeAdapter(tuple[Mapping[str, object], ...])
 _DECODED_JSON: Final = TypeAdapter(object)
@@ -437,20 +480,30 @@ class OpenAIGPTConfig(BaseLLMModelInfo, BaseConfig):
                 )
         return messages, tools
 
+    def translate_developer_role_to_system_role(
+        self,
+        messages: Sequence[AllMessageValues],
+        *,
+        model: str,
+        custom_llm_provider: str | None,
+        api_base: str | None,
+        model_info: object,
+    ) -> Sequence[AllMessageValues]:
+        """
+        A developer message only changes role, except on a model whose deployment or cost
+        map declares `supports_mid_conversation_system` false, where the instructions are
+        placed in one leading system message for its chat template.
+        """
+        return translate_developer_role_for_openai_compatible_endpoint(
+            messages, model=model, custom_llm_provider=custom_llm_provider, api_base=api_base, model_info=model_info
+        )
+
     def _targets_openai_hosted_endpoint(
         self,
         custom_llm_provider: str | None,
         api_base: str | None,
     ) -> bool:
-        if custom_llm_provider != "openai":
-            return False
-        resolved_api_base = api_base or litellm.api_base or os.getenv("OPENAI_BASE_URL") or os.getenv("OPENAI_API_BASE")
-        if not resolved_api_base:
-            return True
-        hostname: Final = urlparse(resolved_api_base).hostname
-        if hostname is None:
-            return True
-        return hostname == "openai.com" or hostname.endswith(".openai.com")
+        return targets_openai_hosted_endpoint(custom_llm_provider, api_base)
 
     def _should_preserve_cache_control_for_endpoint(
         self,

litellm/llms/openai/chat/o_series_transformation.py

@@ -11,7 +11,7 @@ Translations handled by LiteLLM:
 - Logprobs => drop param (if user opts in to dropping param)
 """
 
-from collections.abc import Coroutine
+from collections.abc import Coroutine, Sequence
 from typing import Final, Literal, cast, overload
 
 import litellm
@@ -37,7 +37,15 @@ class OpenAIOSeriesConfig(OpenAIGPTConfig):
     def get_config(cls):
         return super().get_config()
 
-    def translate_developer_role_to_system_role(self, messages: list[AllMessageValues]) -> list[AllMessageValues]:
+    def translate_developer_role_to_system_role(
+        self,
+        messages: Sequence[AllMessageValues],
+        *,
+        model: str,
+        custom_llm_provider: str | None,
+        api_base: str | None,
+        model_info: object,
+    ) -> Sequence[AllMessageValues]:
         """
         O-series models support `developer` role.
         """

litellm/llms/openai/common_utils.py

@@ -104,6 +104,19 @@ def is_openai_backed_api_base(api_base: str) -> bool:
     return hostname is not None and (hostname == _OPENAI_API_HOST or hostname.endswith(f".{_OPENAI_API_HOST}"))
 
 
+_OPENAI_AGENTS_API_RESOURCES: Final = frozenset({"agents", "vaults"})
+
+
+def openai_api_resource(path: str) -> str | None:
+    segments: Final = tuple(segment for segment in path.split("/") if segment)
+    resource_segments: Final = segments[1:] if segments[:1] == ("v1",) else segments
+    return next(iter(resource_segments), None)
+
+
+def is_openai_agents_api_path(path: str) -> bool:
+    return openai_api_resource(path) in _OPENAI_AGENTS_API_RESOURCES
+
+
 class OpenAIError(BaseLLMException):
     def __init__(
         self,

litellm/llms/openai/openai.py

@@ -1,6 +1,6 @@
 import time
 import types
-from collections.abc import AsyncIterator, Callable, Coroutine, Iterable, Iterator, Mapping
+from collections.abc import AsyncIterator, Callable, Coroutine, Iterable, Iterator, Mapping, Sequence
 from typing import TYPE_CHECKING, Final, Literal, Optional, cast
 
 import httpx
@@ -50,7 +50,11 @@ from litellm.utils import (
 from ...types.llms.openai import *
 from ..base import BaseLLM
 from .chat.gpt_reasoning_transformation import OpenAIGPT5Config
-from .chat.gpt_transformation import OpenAIGPTConfig, OpenAIUnknownModelConfig
+from .chat.gpt_transformation import (
+    OpenAIGPTConfig,
+    OpenAIUnknownModelConfig,
+    translate_developer_role_for_openai_compatible_endpoint,
+)
 from .chat.o_series_transformation import OpenAIOSeriesConfig
 from .common_utils import (
     BaseOpenAILLM,
@@ -206,6 +210,19 @@ class OpenAIConfig(BaseConfig):
         else:
             return self._gpt_config_for_model(model).get_supported_openai_params(model=model)
 
+    def translate_developer_role_to_system_role(
+        self,
+        messages: Sequence[AllMessageValues],
+        *,
+        model: str,
+        custom_llm_provider: str | None,
+        api_base: str | None,
+        model_info: object,
+    ) -> Sequence[AllMessageValues]:
+        return translate_developer_role_for_openai_compatible_endpoint(
+            messages, model=model, custom_llm_provider=custom_llm_provider, api_base=api_base, model_info=model_info
+        )
+
     def _gpt_config_for_model(self, model: str) -> OpenAIGPTConfig:
         if type(self) is OpenAIConfig and not OpenAIGPTConfig.is_openai_catalog_model(model):
             return OpenAIUnknownModelConfig()

litellm/llms/openai/responses/instruction_items.py

@@ -0,0 +1,98 @@
+from collections.abc import Mapping
+from types import MappingProxyType
+from typing import Final, Literal, TypeAlias
+
+from openai.types.responses import EasyInputMessageParam, ResponseInputContentParam, ResponseInputItemParam
+
+from litellm.llms.base_llm.base_utils import hoisted_instruction_indices
+from litellm.types.llms.openai import ResponseInputParam
+
+UnfoldedDeveloperRole: TypeAlias = Literal["system", "user"]
+
+_INSTRUCTION_ROLES: Final = frozenset({"system", "developer"})
+
+
+def _role(item: ResponseInputItemParam) -> str | None:
+    match item:
+        case {"role": str(role)}:
+            return role
+        case _:
+            return None
+
+
+def _developer_item_as(item: ResponseInputItemParam, role: UnfoldedDeveloperRole) -> ResponseInputItemParam:
+    if "role" not in item or item["role"] != "developer" or "content" not in item:
+        return item
+    return EasyInputMessageParam(role=role, content=item["content"], type="message")
+
+
+def developer_items_as_system(input: str | ResponseInputParam) -> str | ResponseInputParam:
+    if isinstance(input, str):
+        return input
+    return [_developer_item_as(item, "system") for item in input]
+
+
+def _text_part(part: ResponseInputContentParam) -> str | None:
+    match part:
+        case {"type": "input_text", "text": str(text)}:
+            return text
+        case _:
+            return None
+
+
+def _text_only_content(item: ResponseInputItemParam) -> str | None:
+    match item:
+        case {"role": "system" | "developer", "content": str(text)}:
+            return text
+        case {"role": "system" | "developer", "content": [*parts]}:
+            texts: Final = tuple(map(_text_part, parts))
+            return None if any(text is None for text in texts) else "\n\n".join(text for text in texts if text)
+        case _:
+            return None
+
+
+def _folded_instructions_and_items(
+    items: tuple[ResponseInputItemParam, ...], instructions: str | None, unfolded_developer_role: UnfoldedDeveloperRole
+) -> tuple[str | None, ResponseInputParam]:
+    folded: Final = MappingProxyType(
+        {
+            index: text
+            for index in hoisted_instruction_indices(tuple(map(_role, items)))
+            if (text := _text_only_content(items[index])) is not None
+        }
+    )
+    joined: Final = "\n\n".join(chunk for chunk in (instructions, *folded.values()) if chunk)
+    return (
+        instructions if not folded else joined or None,
+        [_developer_item_as(item, unfolded_developer_role) for index, item in enumerate(items) if index not in folded],
+    )
+
+
+def instruction_items_folded(
+    input: str | ResponseInputParam,
+    params: Mapping[str, object],
+    unfolded_developer_role: UnfoldedDeveloperRole,
+) -> tuple[ResponseInputParam, Mapping[str, object]] | None:
+    """
+    For a model whose chat template takes a single leading system message, the
+    leading system and developer items and every later text developer item folded into
+    top-level `instructions`, the developer items left over sent with
+    `unfolded_developer_role`, or None when the request needs no folding.
+    """
+    instructions: Final = params.get("instructions")
+    if isinstance(input, str) or not isinstance(instructions, str | None):
+        return None
+    items: Final = tuple(input)
+    if not any(_role(item) in _INSTRUCTION_ROLES for item in items):
+        return None
+    folded_instructions, folded_items = _folded_instructions_and_items(items, instructions, unfolded_developer_role)
+    instruction_entries: Final = () if folded_instructions is None else (("instructions", folded_instructions),)
+    return folded_items, MappingProxyType(
+        {
+            key: value
+            for key, value in (
+                *((key, value) for key, value in params.items() if key != "instructions"),
+                *instruction_entries,
+            )
+        }
+    )

litellm/llms/openai/responses/transformation.py

@@ -1,4 +1,4 @@
-from collections.abc import Callable, Collection, Mapping, Sequence
+from collections.abc import Callable, Collection, Iterable, Mapping, Sequence
 from functools import lru_cache
 from types import MappingProxyType
 from typing import TYPE_CHECKING, Any, Final, Protocol, cast, get_type_hints
@@ -10,7 +10,11 @@ from typing_extensions import ReadOnly, TypedDict
 
 import litellm
 from litellm._logging import verbose_logger
-from litellm.integrations.anthropic_cache_control_hook import AnthropicCacheControlHook, targets_openai_api
+from litellm.integrations.anthropic_cache_control_hook import (
+    AnthropicCacheControlHook,
+    openai_prompt_cache_breakpoint_support,
+    targets_openai_api,
+)
 from litellm.litellm_core_utils.core_helpers import normalize_drop_params, process_response_headers
 from litellm.litellm_core_utils.get_model_cost_map import GetModelCostMap
 from litellm.litellm_core_utils.llm_response_utils.convert_dict_to_response import (
@@ -20,6 +24,7 @@ from litellm.litellm_core_utils.prompt_templates.common_utils import (
     cache_control_as_prompt_cache_breakpoint,
     drop_non_python_regex_patterns,
     flatten_combinators_and_drop_non_python_regex_patterns,
+    prompt_cache_breakpoint_for_wire,
     with_prompt_cache_breakpoint,
 )
 from litellm.litellm_core_utils.safe_json_loads import safe_json_loads
@@ -27,6 +32,7 @@ from litellm.litellm_core_utils.url_utils import encode_url_path_segment, join_a
 from litellm.llms.base_llm.responses.transformation import BaseResponsesAPIConfig
 from litellm.llms.openai.chat.gpt_reasoning_transformation import (
     is_gpt_reasoning_series_name,
+    refused_reasoning_effort_warning,
     unsupported_reasoning_effort_message,
 )
 from litellm.responses.litellm_completion_transformation.custom_tools import TOOL_CALL_ITEM_ID_PREFIX_BY_TYPE
@@ -50,6 +56,9 @@ from ..common_utils import OpenAIError
 from ..workload_identity import get_workload_identity_bearer_token, resolve_openai_workload_identity_config
 
 OPENAI_RESPONSES_API_MIN_MAX_OUTPUT_TOKENS: Final = 16
+DROP_UNSUPPORTED_REASONING_WARNING: Final = (
+    "Dropping reasoning=%r for model=%r (drop_params=True) because its model cost map entry lacks supports_reasoning."
+)
 _RAW_RESPONSE_JSON: Final = TypeAdapter(dict[str, object], config=ConfigDict(strict=True))
 
 if TYPE_CHECKING:
@@ -66,6 +75,7 @@ _PROVIDERS_VALIDATING_TOOL_CALL_ITEM_IDS: Final = frozenset({LlmProviders.AZURE,
 _PROVIDERS_REPLAYING_ONLY_THEIR_OWN_REASONING: Final = _PROVIDERS_VALIDATING_TOOL_CALL_ITEM_IDS
 _ROLES_READING_PROMPT_CACHE_BREAKPOINTS: Final = frozenset({"system", "developer", "user"})
 _PART_TYPES_READING_PROMPT_CACHE_BREAKPOINTS: Final = frozenset({"input_text", "input_image", "input_file"})
+_INPUT_PART_LIST_KEYS: Final = frozenset({"content", "output"})
 _COMPACT_TRACKING_PARAMS: Final = frozenset({"metadata", "user"})
 _COMPACT_REJECTED_PARAMS: Final = frozenset(
     {
@@ -93,6 +103,13 @@ _INPUT_ITEM: Final[TypeAdapter[Mapping[str, object]]] = TypeAdapter(Mapping[str,
 _CONTENT_PARTS: Final[TypeAdapter[Sequence[Mapping[str, object]]]] = TypeAdapter(Sequence[Mapping[str, object]])
 
 
+class _DeclaredModel(BaseModel):
+    model_config = ConfigDict(frozen=True, extra="ignore", from_attributes=True)
+
+    base_model: object = None
+    model_info: object = None
+
+
 class _DeploymentHost(BaseModel):
     """`responses()` builds the params from raw kwargs, where callers may spell the host `base_url`."""
 
@@ -216,6 +233,63 @@ def _with_breakpoint_on_last_reading_part(
     return (*parts[:last], marked_last, *parts[last + 1 :])
 
 
+def _json_object_or_none(value: object) -> Mapping[str, object] | None:
+    return value if isinstance(value, Mapping) else None
+
+
+def _json_array_or_empty(value: object) -> Sequence[object]:
+    return value if isinstance(value, (list, tuple)) else ()
+
+
+def _item_with_wire_safe_prompt_cache_breakpoints(item: object, keeps_markers: bool) -> object:
+    fields: Final = _json_object_or_none(item)
+    if fields is None or not _carries_prompt_cache_breakpoint(fields):
+        return item
+    wire_safe_fields: Final = {
+        key: _parts_with_wire_safe_prompt_cache_breakpoints(key, value, keeps_markers)
+        for key, value in fields.items()
+        if key != "prompt_cache_breakpoint"
+    }
+    unchanged: Final = "prompt_cache_breakpoint" not in fields and _same_objects(
+        wire_safe_fields.values(), fields.values()
+    )
+    return item if unchanged else wire_safe_fields
+
+
+def _same_objects(rebuilt: Iterable[object], originals: Iterable[object]) -> bool:
+    return all(new is old for new, old in zip(rebuilt, originals, strict=True))
+
+
+def _carries_prompt_cache_breakpoint(fields: Mapping[str, object]) -> bool:
+    return "prompt_cache_breakpoint" in fields or any(
+        _parts_carry_prompt_cache_breakpoint(fields.get(key)) for key in _INPUT_PART_LIST_KEYS
+    )
+
+
+def _parts_carry_prompt_cache_breakpoint(parts: object) -> bool:
+    return any(isinstance(part, Mapping) and "prompt_cache_breakpoint" in part for part in _json_array_or_empty(parts))
+
+
+def _parts_with_wire_safe_prompt_cache_breakpoints(key: str, value: object, keeps_markers: bool) -> object:
+    if key not in _INPUT_PART_LIST_KEYS or not _parts_carry_prompt_cache_breakpoint(value):
+        return value
+    parts: Final = _json_array_or_empty(value)
+    wire_safe_parts: Final = tuple(_part_with_wire_safe_prompt_cache_breakpoint(part, keeps_markers) for part in parts)
+    return value if _same_objects(wire_safe_parts, parts) else list(wire_safe_parts)
+
+
+def _part_with_wire_safe_prompt_cache_breakpoint(part: object, keeps_markers: bool) -> object:
+    fields: Final = _json_object_or_none(part)
+    if fields is None or "prompt_cache_breakpoint" not in fields:
+        return part
+    marker: Final = fields["prompt_cache_breakpoint"]
+    wire_marker: Final = prompt_cache_breakpoint_for_wire(marker, drop_params=True) if keeps_markers else None
+    if keeps_markers and wire_marker == marker:
+        return part
+    unmarked: Final = {key: value for key, value in fields.items() if key != "prompt_cache_breakpoint"}
+    return with_prompt_cache_breakpoint(unmarked, wire_marker)
+
+
 class _ReasoningSupportEntry(LiteLLMBaseModel):
     litellm_provider: str | None = None
     supports_reasoning: bool | None = None
@@ -347,12 +421,19 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
             return declared
         return info["key"] in _bundled_openai_reasoning_models()
 
+    @staticmethod
+    def _is_known_non_reasoning_model(model: str) -> bool:
+        from litellm.utils import is_known_non_reasoning_model
+
+        return is_known_non_reasoning_model(model.split("/")[-1], LlmProviders.OPENAI.value)
+
     @staticmethod
     def _requests_reasoning_effort(reasoning: object) -> bool:
-        effort: Final = (
-            reasoning.get("effort") if isinstance(reasoning, Mapping) else getattr(reasoning, "effort", None)
-        )
-        return effort is not None
+        return OpenAIResponsesAPIConfig._requested_effort(reasoning) is not None
+
+    @staticmethod
+    def _requested_effort(reasoning: object) -> object:
+        return reasoning.get("effort") if isinstance(reasoning, Mapping) else getattr(reasoning, "effort", None)
 
     @staticmethod
     def _enforce_min_max_output_tokens(max_output_tokens: "int | None") -> "int | None":
@@ -394,17 +475,20 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
     ) -> dict:
         return dict(
             self.map_openai_params_for_deployment(
-                response_api_optional_params, model, drop_params, declared_reasoning_efforts=()
+                response_api_optional_params, model, drop_params, declared_reasoning_efforts=None
             )
         )
 
+    def honors_declared_reasoning_efforts(self) -> bool:
+        return type(self).map_openai_params is OpenAIResponsesAPIConfig.map_openai_params
+
     def map_openai_params_for_deployment(
         self,
         response_api_optional_params: ResponsesAPIOptionalRequestParams,
         model: str,
         drop_params: bool,
         *,
-        declared_reasoning_efforts: Collection[str],
+        declared_reasoning_efforts: Collection[str] | None,
     ) -> Mapping[str, object]:
         """No mapping applied since inputs are in OpenAI spec already.
 
@@ -424,7 +508,9 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
             and not self._supports_reasoning_param(model=model)
         ):
             if drop_params or litellm.drop_params:
-                params.pop("reasoning", None)
+                dropped_reasoning: Final = params.pop("reasoning", None)
+                if self._requested_effort(dropped_reasoning) != "none" or not self._is_known_non_reasoning_model(model):
+                    verbose_logger.warning(DROP_UNSUPPORTED_REASONING_WARNING, dropped_reasoning, model)
             else:
                 raise litellm.UnsupportedParamsError(
                     message=(
@@ -442,7 +528,7 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
             if (
                 isinstance(reasoning, dict)
                 and isinstance(requested_effort, str)
-                and requested_effort not in declared_reasoning_efforts
+                and requested_effort not in (declared_reasoning_efforts or ())
                 and self._effort_level_is_disabled(lookup_name, requested_effort)
             ):
                 if not (drop_params or litellm.drop_params):
@@ -450,6 +536,15 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
                         message=unsupported_reasoning_effort_message("reasoning.effort", requested_effort, model),
                         status_code=400,
                     )
+                verbose_logger.warning(
+                    refused_reasoning_effort_warning(
+                        suggest_declared_levels=declared_reasoning_efforts is not None, suggest_base_model=False
+                    ),
+                    "reasoning.effort",
+                    requested_effort,
+                    model,
+                    lookup_name,
+                )
                 remaining: Final = reasoning.copy()
                 remaining.pop("effort", None)
                 if remaining:
@@ -544,12 +639,11 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
         validated_input: Final = self._validate_input_param(
             self._drop_bridge_minted_reasoning_items(input), drops_set_created_by=drops_params(litellm_params)
         )
-        marked_input: Final = (
-            self._input_with_prompt_cache_breakpoints(validated_input)
-            if self._reads_prompt_cache_breakpoints(
-                model=model, litellm_params=litellm_params, prompt_cache_options=prompt_cache_options
-            )
-            else validated_input
+        reads_breakpoints: Final = self._reads_prompt_cache_breakpoints(
+            model=model, litellm_params=litellm_params, prompt_cache_options=prompt_cache_options
+        )
+        marker_safe_input: Final = self._input_with_wire_safe_prompt_cache_breakpoints(
+            model=model, input=validated_input, litellm_params=litellm_params, reads_breakpoints=reads_breakpoints
         )
         maps_anthropic_tool_search: Final = self._targets_openai_api(litellm_params) and any(
             is_anthropic_tool_search_chat_tool(tool) for tool in tools or ()
@@ -559,20 +653,71 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
             if maps_anthropic_tool_search
             else tools
         )
-        replayable_input: Final = (
-            input_without_unanswered_anthropic_tool_search_calls(marked_input)
-            if maps_anthropic_tool_search
-            else marked_input
-        )
         stripped_input, stripped_tools = self.remove_cache_control_flag_from_input_and_tools(
-            model=model, input=replayable_input, tools=mapped_tools
+            model=model,
+            input=self._replayable_input(marker_safe_input, reads_breakpoints, maps_anthropic_tool_search),
+            tools=mapped_tools,
         )
+        if marker_safe_input is not validated_input:
+            self._strip_caller_input_as_if_unmarked(
+                model=model,
+                caller_input=validated_input,
+                reads_breakpoints=reads_breakpoints,
+                maps_anthropic_tool_search=maps_anthropic_tool_search,
+            )
         object_schema_tools: Final = self._tools_with_object_parameters(model=model, tools=stripped_tools)
         sanitized_tools: Final = self._sanitized_tool_schemas_for_openai(
             model=model, tools=object_schema_tools, litellm_params=litellm_params
         )
         return self._drop_foreign_tool_call_item_ids(stripped_input), sanitized_tools
 
+    def _replayable_input(
+        self, input: str | ResponseInputParam, reads_breakpoints: bool, maps_anthropic_tool_search: bool
+    ) -> str | ResponseInputParam:
+        marked_input: Final = self._input_with_prompt_cache_breakpoints(input) if reads_breakpoints else input
+        if not maps_anthropic_tool_search:
+            return marked_input
+        return input_without_unanswered_anthropic_tool_search_calls(marked_input)
+
+    def _strip_caller_input_as_if_unmarked(
+        self,
+        model: str,
+        caller_input: str | ResponseInputParam,
+        reads_breakpoints: bool,
+        maps_anthropic_tool_search: bool,
+    ) -> None:
+        self.remove_cache_control_flag_from_input_and_tools(
+            model=model, input=self._replayable_input(caller_input, reads_breakpoints, maps_anthropic_tool_search)
+        )
+
+    def _input_with_wire_safe_prompt_cache_breakpoints(
+        self,
+        model: str,
+        input: str | ResponseInputParam,
+        litellm_params: GenericLiteLLMParams,
+        reads_breakpoints: bool,
+    ) -> str | ResponseInputParam:
+        if not isinstance(input, list) or not drops_params(litellm_params):
+            return input
+        keeps_markers: Final = reads_breakpoints or self._keeps_prompt_cache_breakpoints(
+            model=model, litellm_params=litellm_params
+        )
+        wire_safe_items: Final = tuple(
+            _item_with_wire_safe_prompt_cache_breakpoints(item, keeps_markers) for item in input
+        )
+        if _same_objects(wire_safe_items, input):
+            return input
+        wire_safe_input: Final = list(wire_safe_items)
+        return cast("ResponseInputParam", wire_safe_input)  # cast-ok: items keep their shape, minus a rejected marker
+
+    def _keeps_prompt_cache_breakpoints(self, model: str, litellm_params: GenericLiteLLMParams) -> bool:
+        provider: Final = _provider_name(self.custom_llm_provider)
+        base_model: Final = self._declared_base_model(model=model, litellm_params=litellm_params)
+        known_support: Final = (
+            openai_prompt_cache_breakpoint_support(name, provider) for name in dict.fromkeys((base_model, model))
+        )
+        return next((support for support in known_support if support is not None), None) is not False
+
     def _targets_openai_api(self, litellm_params: GenericLiteLLMParams) -> bool:
         host: Final = _DeploymentHost.model_validate(litellm_params)
         return self.custom_llm_provider == LlmProviders.OPENAI and targets_openai_api(host.api_base or host.base_url)
@@ -688,14 +833,15 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
 
     @staticmethod
     def _without_foreign_tool_call_item_id(item: object) -> object:
-        if not isinstance(item, dict):
+        fields: Final = _json_object_or_none(item)
+        if fields is None:
             return item
-        item_type: Final = item.get("type")
-        item_id: Final = item.get("id")
+        item_type: Final = fields.get("type")
+        item_id: Final = fields.get("id")
         genuine_prefix: Final = TOOL_CALL_ITEM_ID_PREFIX_BY_TYPE.get(item_type) if isinstance(item_type, str) else None
         if genuine_prefix is None or not isinstance(item_id, str) or item_id.startswith(genuine_prefix):
             return item
-        return {key: value for key, value in item.items() if key != "id"}
+        return {key: value for key, value in fields.items() if key != "id"}
 
     def _sanitized_tool_schemas_for_openai(
         self,
@@ -739,6 +885,13 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
         base_model: Final[object] = model_info.get("base_model") if isinstance(model_info, dict) else None
         return base_model if isinstance(base_model, str) and base_model else model
 
+    @staticmethod
+    def _declared_base_model(model: str, litellm_params: GenericLiteLLMParams) -> str:
+        declared: Final = _DeclaredModel.model_validate(litellm_params)
+        model_info: Final = _json_object_or_none(declared.model_info)
+        candidates: Final = (declared.base_model, None if model_info is None else model_info.get("base_model"))
+        return next((name for name in candidates if isinstance(name, str) and name), model)
+
     @staticmethod
     def _sanitized_tool_entry(
         entry: Mapping[str, object],
@@ -1094,7 +1247,7 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
         api_base: str,
         litellm_params: GenericLiteLLMParams,
         headers: dict,
-    ) -> tuple[str, dict]:
+    ) -> tuple[str, dict[str, str]]:
         """
         Transform the get response API request into a URL and data
 
@@ -1103,7 +1256,7 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
         """
         encoded_response_id: Final = encode_url_path_segment(response_id, field_name="response_id")
         url: Final = f"{api_base}/{encoded_response_id}"
-        data: Final[dict] = {}
+        data: Final[dict[str, str]] = {}
         return url, data
 
     def transform_get_response_api_response(

litellm/llms/openai_like/messages/transformation.py

@@ -37,6 +37,9 @@ class OpenAILikeAnthropicMessagesConfig(AnthropicMessagesConfig):
     def is_native_messages_route(self) -> bool:
         return False
 
+    def hoists_wrapped_defer_loading(self, model: str) -> bool:
+        return False
+
     def validate_anthropic_messages_environment(
         self,
         headers: dict[str, str],

litellm/llms/openai_like/providers.json

@@ -270,6 +270,13 @@
       "/v1/responses"
     ]
   },
+  "kyma": {
+    "base_url": "https://kymaapi.com/v1",
+    "api_key_env": "KYMA_API_KEY",
+    "provider_info": {
+      "supports_function_calling": true
+    }
+  },
   "streamlake": {
     "base_url": "https://vanchin.streamlake.ai/api/gateway/v1/endpoints",
     "api_key_env": "STREAMLAKE_API_KEY",

litellm/llms/vertex_ai/context_caching/vertex_ai_context_caching.py

@@ -1,4 +1,5 @@
 from collections.abc import Sequence
+from dataclasses import dataclass
 from typing import Final, Literal
 
 import httpx
@@ -8,7 +9,6 @@ from typing_extensions import NotRequired, ReadOnly, TypedDict
 import litellm
 from litellm._logging import verbose_logger
 from litellm.caching.caching import Cache, LiteLLMCacheType
-from litellm.constants import MINIMUM_PROMPT_CACHE_TOKEN_COUNT
 from litellm.litellm_core_utils.litellm_logging import Logging
 from litellm.llms.custom_httpx.http_handler import (
     AsyncHTTPHandler,
@@ -16,10 +16,11 @@ from litellm.llms.custom_httpx.http_handler import (
     get_async_httpx_client,
 )
 from litellm.llms.openai.openai import AllMessageValues
-from litellm.utils import is_prompt_caching_valid_prompt
+from litellm.types.llms.openai import ChatCompletionToolParam
+from litellm.utils import get_prompt_cache_min_tokens, is_prompt_caching_valid_prompt
 
 from ..common_utils import VertexAIError, get_vertex_base_url
-from ..vertex_llm_base import VertexBase
+from ..vertex_llm_base import VertexBase, publisher_model_location
 from .known_cached_contents import KnownCachedContents, known_cached_contents
 from .transformation import (
     cached_messages_end_on_supported_turn,
@@ -65,6 +66,19 @@ class _CreatedCachedContentRoot(RootModel[_CreatedCachedContent]):
     model_config = ConfigDict(frozen=True, hide_input_in_errors=True)
 
 
+_TOOL_PARAMS: Final = ("tools", "tool_choice")
+
+
+def _sync_cache_client(timeout: float | httpx.Timeout | None) -> HTTPHandler:
+    if timeout is None:
+        return HTTPHandler()
+    return HTTPHandler(timeout=httpx.Timeout(timeout) if isinstance(timeout, (float, int)) else timeout)
+
+
+class _CachedContentsNotServedError(VertexAIError):
+    pass
+
+
 def _record_cache_creation_tokens(logging_obj: Logging, created: _CreatedCachedContent) -> None:
     tokens: Final = (created.get("usageMetadata") or {}).get("totalTokenCount") or 0
     logging_obj.vertex_ai_cache_creation_input_tokens = max(tokens, 0)  # rebind-ok: set provider count
@@ -78,6 +92,61 @@ def _wait_seconds(timeout: float | httpx.Timeout | None) -> float | None:
             return timeout
 
 
+@dataclass(frozen=True, slots=True)
+class _CacheTarget:
+    model: str
+    project: str | None
+    location: str | None
+
+
+def _cache_target(
+    custom_llm_provider: Literal["vertex_ai", "vertex_ai_beta", "gemini"],
+    api_base: str | None,
+    model: str,
+    vertex_project: str | None,
+    vertex_location: str | None,
+) -> _CacheTarget:
+    model_location: Final = (
+        None if custom_llm_provider == "gemini" or api_base is None else publisher_model_location(api_base)
+    )
+    if model_location is None:
+        return _CacheTarget(model=model, project=vertex_project, location=vertex_location)
+    return _CacheTarget(model=model_location.model, project=model_location.project, location=model_location.location)
+
+
+def _prompt_cache_min_tokens(models: Sequence[str], custom_llm_provider: str) -> int:
+    return max(
+        get_prompt_cache_min_tokens(
+            model if model.startswith(custom_llm_provider) else f"{custom_llm_provider}/{model}"
+        )
+        for model in models
+    )
+
+
+def _context_caching_skip_reason(
+    cached_messages: Sequence[AllMessageValues],
+    tools: Sequence[ChatCompletionToolParam] | None,
+    deployment_model: str,
+    cache_model: str,
+    custom_llm_provider: Literal["vertex_ai", "vertex_ai_beta", "gemini"],
+) -> str | None:
+    if not cached_messages_end_on_supported_turn(cached_messages):
+        return (
+            "Vertex AI context caching: cached message block ends on a model turn once "
+            "system messages are extracted, which the cachedContents API rejects."
+        )
+    min_tokens: Final = _prompt_cache_min_tokens((deployment_model, cache_model), custom_llm_provider)
+    if not is_prompt_caching_valid_prompt(
+        model=cache_model,
+        messages=list(cached_messages),
+        tools=None if tools is None else list(tools),
+        custom_llm_provider=custom_llm_provider,
+        min_token_count=min_tokens,
+    ):
+        return f"Vertex AI context caching: cached content is below minimum token count ({min_tokens})."
+    return None
+
+
 class ContextCachingEndpoints(VertexBase):
     """
     Covers context caching endpoints for Vertex AI + Google AI Studio
@@ -135,6 +204,7 @@ class ContextCachingEndpoints(VertexBase):
             vertex_project=vertex_project,
             vertex_location=vertex_location,
             vertex_api_version=("v1beta1" if custom_llm_provider == "vertex_ai_beta" else "v1"),
+            collection_endpoint=True,
         )
 
     def check_cache(
@@ -206,6 +276,8 @@ class ContextCachingEndpoints(VertexBase):
             except httpx.HTTPStatusError as e:
                 if e.response.status_code == 403:
                     return None
+                if e.response.status_code == 404 and custom_llm_provider == "gemini":
+                    raise _CachedContentsNotServedError(status_code=404, message=e.response.text)
                 raise VertexAIError(status_code=e.response.status_code, message=e.response.text)
             except Exception as e:
                 raise VertexAIError(status_code=500, message=str(e))
@@ -305,6 +377,8 @@ class ContextCachingEndpoints(VertexBase):
             except httpx.HTTPStatusError as e:
                 if e.response.status_code == 403:
                     return None
+                if e.response.status_code == 404 and custom_llm_provider == "gemini":
+                    raise _CachedContentsNotServedError(status_code=404, message=e.response.text)
                 raise VertexAIError(status_code=e.response.status_code, message=e.response.text)
             except Exception as e:
                 raise VertexAIError(status_code=500, message=str(e))
@@ -370,41 +444,25 @@ class ContextCachingEndpoints(VertexBase):
         if len(cached_messages) == 0:
             return messages, optional_params, None
 
-        if not cached_messages_end_on_supported_turn(cached_messages):
-            verbose_logger.debug(
-                "Vertex AI context caching: cached message block ends on a model turn once "
-                "system messages are extracted, which the cachedContents API rejects. "
-                "Skipping context caching."
-            )
+        target: Final = _cache_target(custom_llm_provider, api_base, model, vertex_project, vertex_location)
+        tools: Final = optional_params.get("tools")
+        tool_choice: Final = optional_params.get("tool_choice")
+        skip_reason: Final = _context_caching_skip_reason(
+            cached_messages, tools, model, target.model, custom_llm_provider
+        )
+        if skip_reason is not None:
+            verbose_logger.debug("%s Skipping context caching.", skip_reason)
             return messages, optional_params, None
 
-        # Gemini requires a minimum of 1024 tokens for context caching.
-        # Skip caching if the cached content is too small to avoid API errors.
-        if not is_prompt_caching_valid_prompt(
-            model=model,
-            messages=cached_messages,
-            tools=optional_params.get("tools"),
-            custom_llm_provider=custom_llm_provider,
-        ):
-            verbose_logger.debug(
-                "Vertex AI context caching: cached content is below minimum token "
-                "count (%d). Skipping context caching.",
-                MINIMUM_PROMPT_CACHE_TOKEN_COUNT,
-            )
-            return messages, optional_params, None
-
-        tools: Final = optional_params.pop("tools", None)
-        tool_choice: Final = optional_params.pop("tool_choice", None)
-
         ## AUTHORIZATION ##
         token, url = self._get_token_and_url_context_caching(
             gemini_api_key=api_key,
             custom_llm_provider=custom_llm_provider,
             api_base=api_base,
-            vertex_project=vertex_project,
-            vertex_location=vertex_location,
+            vertex_project=target.project,
+            vertex_location=target.location,
             vertex_auth_header=vertex_auth_header,
-            model=model,
+            model=target.model,
         )
 
         headers: Final = {
@@ -417,44 +475,42 @@ class ContextCachingEndpoints(VertexBase):
         if extra_headers is not None:
             headers.update(extra_headers)
 
-        if client is None or not isinstance(client, HTTPHandler):
-            _params: Final = {}
-            if timeout is not None:
-                if isinstance(timeout, float) or isinstance(timeout, int):
-                    timeout = httpx.Timeout(timeout)
-                _params["timeout"] = timeout
-            client = HTTPHandler(**_params)
-        else:
-            client = client
+        http_client: Final = client if isinstance(client, HTTPHandler) else _sync_cache_client(timeout)
 
         ## CHECK IF CACHED ALREADY
         generated_cache_key: Final = local_cache_obj.get_cache_key(
-            messages=cached_messages, tools=tools, tool_choice=tool_choice, model=model
-        )
-        google_cache_name: Final = self.check_cache(
-            cache_key=generated_cache_key,
-            client=client,
-            headers=headers,
-            api_key=api_key,
-            api_base=api_base,
-            logging_obj=logging_obj,
-            custom_llm_provider=custom_llm_provider,
-            vertex_project=vertex_project,
-            vertex_location=vertex_location,
-            vertex_auth_header=vertex_auth_header,
-            model=model,
+            messages=cached_messages, tools=tools, tool_choice=tool_choice, model=target.model
         )
+        try:
+            google_cache_name: Final = self.check_cache(
+                cache_key=generated_cache_key,
+                client=http_client,
+                headers=headers,
+                api_key=api_key,
+                api_base=api_base,
+                logging_obj=logging_obj,
+                custom_llm_provider=custom_llm_provider,
+                vertex_project=target.project,
+                vertex_location=target.location,
+                vertex_auth_header=vertex_auth_header,
+                model=target.model,
+            )
+        except _CachedContentsNotServedError:
+            verbose_logger.debug("Gemini context caching: %s answers 404. Skipping context caching.", url)
+            return messages, optional_params, None
+        for tool_param in _TOOL_PARAMS:
+            optional_params.pop(tool_param, None)
         if google_cache_name:
             return non_cached_messages, optional_params, google_cache_name
 
         ## TRANSFORM REQUEST
         cached_content_request_body: Final = transform_openai_messages_to_gemini_context_caching(
-            model=model,
+            model=target.model,
             messages=cached_messages,
             cache_key=generated_cache_key,
             custom_llm_provider=custom_llm_provider,
-            vertex_project=vertex_project,
-            vertex_location=vertex_location,
+            vertex_project=target.project,
+            vertex_location=target.location,
         )
 
         cached_content_request_body["tools"] = tools
@@ -473,7 +529,7 @@ class ContextCachingEndpoints(VertexBase):
         )
 
         try:
-            response: Final = client.post(
+            response: Final = http_client.post(
                 url=url,
                 headers=headers,
                 json=cached_content_request_body,
@@ -489,7 +545,7 @@ class ContextCachingEndpoints(VertexBase):
         _record_cache_creation_tokens(logging_obj, raw_response_cached)
         self._known_caches.remember(
             KnownCachedContents.scope(
-                url, custom_llm_provider, headers, vertex_project=vertex_project, vertex_location=vertex_location
+                url, custom_llm_provider, headers, vertex_project=target.project, vertex_location=target.location
             ),
             generated_cache_key,
             raw_response_cached.get("name"),
@@ -536,41 +592,25 @@ class ContextCachingEndpoints(VertexBase):
         if len(cached_messages) == 0:
             return messages, optional_params, None
 
-        if not cached_messages_end_on_supported_turn(cached_messages):
-            verbose_logger.debug(
-                "Vertex AI context caching: cached message block ends on a model turn once "
-                "system messages are extracted, which the cachedContents API rejects. "
-                "Skipping context caching."
-            )
+        target: Final = _cache_target(custom_llm_provider, api_base, model, vertex_project, vertex_location)
+        tools: Final = optional_params.get("tools")
+        tool_choice: Final = optional_params.get("tool_choice")
+        skip_reason: Final = _context_caching_skip_reason(
+            cached_messages, tools, model, target.model, custom_llm_provider
+        )
+        if skip_reason is not None:
+            verbose_logger.debug("%s Skipping context caching.", skip_reason)
             return messages, optional_params, None
 
-        # Gemini requires a minimum of 1024 tokens for context caching.
-        # Skip caching if the cached content is too small to avoid API errors.
-        if not is_prompt_caching_valid_prompt(
-            model=model,
-            messages=cached_messages,
-            tools=optional_params.get("tools"),
-            custom_llm_provider=custom_llm_provider,
-        ):
-            verbose_logger.debug(
-                "Vertex AI context caching: cached content is below minimum token "
-                "count (%d). Skipping context caching.",
-                MINIMUM_PROMPT_CACHE_TOKEN_COUNT,
-            )
-            return messages, optional_params, None
-
-        tools: Final = optional_params.pop("tools", None)
-        tool_choice: Final = optional_params.pop("tool_choice", None)
-
         ## AUTHORIZATION ##
         token, url = self._get_token_and_url_context_caching(
             gemini_api_key=api_key,
             custom_llm_provider=custom_llm_provider,
             api_base=api_base,
-            vertex_project=vertex_project,
-            vertex_location=vertex_location,
+            vertex_project=target.project,
+            vertex_location=target.location,
             vertex_auth_header=vertex_auth_header,
-            model=model,
+            model=target.model,
         )
 
         headers: Final = {
@@ -590,10 +630,10 @@ class ContextCachingEndpoints(VertexBase):
 
         ## CHECK IF CACHED ALREADY
         generated_cache_key: Final = local_cache_obj.get_cache_key(
-            messages=cached_messages, tools=tools, tool_choice=tool_choice, model=model
+            messages=cached_messages, tools=tools, tool_choice=tool_choice, model=target.model
         )
         scope: Final = KnownCachedContents.scope(
-            url, custom_llm_provider, headers, vertex_project=vertex_project, vertex_location=vertex_location
+            url, custom_llm_provider, headers, vertex_project=target.project, vertex_location=target.location
         )
 
         async def lookup_or_create() -> str | None:
@@ -605,10 +645,10 @@ class ContextCachingEndpoints(VertexBase):
                 api_base=api_base,
                 logging_obj=logging_obj,
                 custom_llm_provider=custom_llm_provider,
-                vertex_project=vertex_project,
-                vertex_location=vertex_location,
+                vertex_project=target.project,
+                vertex_location=target.location,
                 vertex_auth_header=vertex_auth_header,
-                model=model,
+                model=target.model,
             )
 
             if google_cache_name:
@@ -616,12 +656,12 @@ class ContextCachingEndpoints(VertexBase):
 
             ## TRANSFORM REQUEST
             cached_content_request_body: Final = transform_openai_messages_to_gemini_context_caching(
-                model=model,
+                model=target.model,
                 messages=cached_messages,
                 cache_key=generated_cache_key,
                 custom_llm_provider=custom_llm_provider,
-                vertex_project=vertex_project,
-                vertex_location=vertex_location,
+                vertex_project=target.project,
+                vertex_location=target.location,
             )
 
             cached_content_request_body["tools"] = tools
@@ -659,9 +699,15 @@ class ContextCachingEndpoints(VertexBase):
             )
             return raw_response_cached.get("name")
 
-        cache_name: Final = await self._known_caches.single_flight(
-            scope, generated_cache_key, lookup_or_create, _wait_seconds(timeout)
-        )
+        try:
+            cache_name: Final = await self._known_caches.single_flight(
+                scope, generated_cache_key, lookup_or_create, _wait_seconds(timeout)
+            )
+        except _CachedContentsNotServedError:
+            verbose_logger.debug("Gemini context caching: %s answers 404. Skipping context caching.", url)
+            return messages, optional_params, None
+        for tool_param in _TOOL_PARAMS:
+            optional_params.pop(tool_param, None)
         return non_cached_messages, optional_params, cache_name
 
     def get_cache(self):

litellm/llms/vertex_ai/vertex_ai_partner_models/anthropic/messages/transformation.py

@@ -211,6 +211,12 @@ class VertexAIPartnerModelsAnthropicMessagesConfig(AnthropicMessagesConfig, Vert
         if messages_carry_output_config(messages):
             beta_values.add(ANTHROPIC_BETA_HEADER_VALUES.PER_TURN_CONTROL_2026_07_01.value)
 
+        beta_values.update(
+            anthropic_model_info.thinking_betas(
+                optional_params.get("thinking"),  # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType]  # optional_params is the base config's bare dict
+            )
+        )
+
         if beta_values:
             request_headers["anthropic-beta"] = ",".join(beta_values)
 
@@ -245,7 +251,9 @@ class VertexAIPartnerModelsAnthropicMessagesConfig(AnthropicMessagesConfig, Vert
             headers=headers,
         )
 
-        self._normalize_system_role_messages(anthropic_messages_request, model=model)
+        self._normalize_system_role_messages(
+            anthropic_messages_request, model=model, drop_params=getattr(litellm_params, "drop_params", None)
+        )
 
         self._remove_scope_from_cache_control(anthropic_messages_request)
 

litellm/llms/vertex_ai/vertex_ai_partner_models/anthropic/transformation.py

@@ -145,6 +145,7 @@ class VertexAIAnthropicConfig(AnthropicConfig):
             file_id_used=self.is_file_id_used(messages),
             mcp_server_used=self.is_mcp_server_used(optional_params.get("mcp_servers")),
             custom_llm_provider="vertex_ai",
+            thinking=optional_params.get("thinking"),  # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType]  # optional_params is the base config's bare dict
         )
 
         beta_set: Final = set(auto_betas)

litellm/llms/vertex_ai/vertex_ai_partner_models/count_tokens/handler.py

@@ -6,9 +6,12 @@ Unlike Gemini models which use Google's token counting API, partner models use
 their respective publisher-specific count-tokens endpoints.
 """
 
+from collections.abc import Mapping
 from typing import Any, Final
 
+from litellm.litellm_core_utils.core_helpers import drops_unsupported_params
 from litellm.litellm_core_utils.url_utils import join_api_base
+from litellm.llms.anthropic.common_utils import OBJECT_LIST_ADAPTER, is_claude_model_target, tools_without_custom_field
 from litellm.llms.custom_httpx.http_handler import get_async_httpx_client
 from litellm.llms.vertex_ai.common_utils import get_vertex_base_url
 from litellm.llms.vertex_ai.vertex_llm_base import VertexBase
@@ -20,6 +23,22 @@ _COUNT_TOKENS_PUBLISHERS: Final = (
 )
 
 
+def count_tokens_request_body(
+    request_data: Mapping[str, object], model: str, drop_params: object
+) -> Mapping[str, object]:
+    tools: Final = request_data.get("tools")
+    if not isinstance(tools, list):
+        return request_data
+    return {
+        **request_data,
+        "tools": tools_without_custom_field(
+            OBJECT_LIST_ADAPTER.validate_python(tools),
+            hoist_defer_loading=is_claude_model_target("vertex_ai", model),
+            drop_custom=drops_unsupported_params(drop_params),
+        ),
+    }
+
+
 def count_tokens_publisher(model: str) -> str | None:
     return next((publisher for marker, publisher in _COUNT_TOKENS_PUBLISHERS if marker in model), None)
 
@@ -169,7 +188,7 @@ class VertexAIPartnerModelsTokenCounter(VertexBase):
         response: Final = await async_client.post(
             endpoint_url,
             headers=headers,
-            json=request_data,
+            json=dict(count_tokens_request_body(request_data, model, litellm_params.get("drop_params"))),
             timeout=30.0,
         )
 

litellm/llms/vertex_ai/vertex_llm_base.py

@@ -6,9 +6,11 @@ Handles Authentication and generating request urls for Vertex AI and Google AI S
 
 import asyncio
 import json
+import re
 import threading
 from collections import OrderedDict
 from collections.abc import Mapping
+from dataclasses import dataclass
 from typing import TYPE_CHECKING, Any, Final, Literal, Protocol
 from urllib.parse import urlparse
 
@@ -46,6 +48,31 @@ def graft_default_vertex_path(api_base: str, default_url: str) -> str:
 
 _graft_default_vertex_path = graft_default_vertex_path
 
+
+_PUBLISHER_MODEL_LOCATION: Final = re.compile(r".*/projects/[^/]+/locations/[^/]+(?=/publishers/[^/]+/models/[^/]+/?$)")
+
+
+@dataclass(frozen=True, slots=True)
+class PublisherModelLocation:
+    parent_url: str
+    project: str
+    location: str
+    model: str
+
+
+def publisher_model_location(api_base: str) -> PublisherModelLocation | None:
+    match: Final = _PUBLISHER_MODEL_LOCATION.match(api_base)
+    if match is None:
+        return None
+    _, _, project, _, location = match.group().rsplit("/", 4)
+    return PublisherModelLocation(
+        parent_url=match.group(),
+        project=project,
+        location=location,
+        model=api_base.rstrip("/").rsplit("/", 1)[-1],
+    )
+
+
 GOOGLE_IMPORT_ERROR_MESSAGE: Final = (
     "Google Cloud SDK not found. Install it with: pip install 'litellm[google]' or pip install google-cloud-aiplatform"
 )
@@ -637,15 +664,18 @@ class VertexBase:
         vertex_location: str | None = None,
         vertex_api_version: Literal["v1", "v1beta1"] | None = None,
         use_psc_endpoint_format: bool = False,
+        collection_endpoint: bool = False,
     ) -> tuple[str | None, str]:
         """
         for cloudflare ai gateway - https://github.com/BerriAI/litellm/issues/4317
 
         Handles custom api_base for:
-        1. Gemini (Google AI Studio) - constructs /models/{model}:{endpoint}
+        1. Gemini (Google AI Studio) - constructs /models/{model}:{endpoint}, or {api_base}/{endpoint}
+           for a collection endpoint
         2. Vertex AI with standard proxies - grafts the default vertex URL path onto the
            api_base when its path is empty or only an API version (/v1, /v1beta1);
-           otherwise constructs {api_base}:{endpoint}
+           otherwise constructs {api_base}:{endpoint}, or, for a collection endpoint under an api_base
+           naming a publisher model, the collection under that model's projects/{project}/locations/{location}
         3. Vertex AI with PSC endpoints - constructs full path structure
            {api_base}/v1/projects/{project}/locations/{location}/endpoints/{model}:{endpoint}
            (only when use_psc_endpoint_format=True)
@@ -653,22 +683,22 @@ class VertexBase:
         Args:
             use_psc_endpoint_format: If True, constructs PSC endpoint URL format.
                                      If False (default), uses api_base as-is and appends :{endpoint}
+            collection_endpoint: True when `endpoint` names a collection resource such as
+                                 cachedContents rather than a model action such as generateContent
 
         ## Returns
         - (auth_header, url) - Tuple[Optional[str], str]
         """
         if api_base:
             if custom_llm_provider == "gemini":
-                # For Gemini (Google AI Studio), construct the full path like other providers
-                if model is None:
+                if model is None and not collection_endpoint:
                     raise ValueError("Model parameter is required for Gemini custom API base URLs")
-                url = join_api_base(api_base, f"models/{model}:{endpoint}")
+                url = join_api_base(api_base, endpoint if collection_endpoint else f"models/{model}:{endpoint}")
                 if gemini_api_key is None:
                     raise ValueError(
                         "Missing Gemini API key. Set the GEMINI_API_KEY or GOOGLE_API_KEY environment variable."
                     )
-                if gemini_api_key is not None:
-                    auth_header = {"x-goog-api-key": gemini_api_key}
+                auth_header = {"x-goog-api-key": gemini_api_key}
             else:
                 # For Vertex AI
                 if use_psc_endpoint_format:
@@ -692,7 +722,12 @@ class VertexBase:
                 elif urlparse(api_base).path.rstrip("/") in ("/v1", "/v1beta1") and "/projects/" in urlparse(url).path:
                     url = graft_default_vertex_path(api_base=api_base, default_url=url)
                 else:
-                    url = f"{api_base.rstrip('/')}:{endpoint}"
+                    model_location: Final = publisher_model_location(api_base) if collection_endpoint else None
+                    url = (
+                        f"{api_base.rstrip('/')}:{endpoint}"
+                        if model_location is None
+                        else join_api_base(model_location.parent_url, endpoint)
+                    )
             if stream is True:
                 parsed_stream_url: Final = urlparse(url)
                 stream_query: Final = f"{parsed_stream_url.query}&alt=sse" if parsed_stream_url.query else "alt=sse"

litellm/llms/volcengine/responses/transformation.py

@@ -313,10 +313,10 @@ class VolcEngineResponsesAPIConfig(OpenAIResponsesAPIConfig):
         api_base: str,
         litellm_params: GenericLiteLLMParams,
         headers: dict,
-    ) -> tuple[str, dict]:
+    ) -> tuple[str, dict[str, str]]:
         encoded_response_id: Final = encode_url_path_segment(response_id, field_name="response_id")
         url: Final = f"{api_base}/{encoded_response_id}"
-        data: Final[dict] = {}
+        data: Final[dict[str, str]] = {}
         return url, data
 
     def transform_get_response_api_response(

litellm/main.py

@@ -155,6 +155,7 @@ from litellm.types.litellm_params import ControlOptions, RetryStrategy
 from litellm.types.llms.openai import AllMessageValues
 from litellm.types.router import GenericLiteLLMParams
 from litellm.types.utils import (
+    AgenticLoopParams,
     BaseUrlRoutingHint,
     CustomPricingLiteLLMParams,
     ModelResponseStream,
@@ -234,6 +235,7 @@ from .litellm_core_utils.streaming_chunk_builder_utils import (
     ChunkProcessor as ChunkProcessor,  # noqa: PLC0414  # public re-export
 )
 from .llms.anthropic.chat import AnthropicChatCompletion
+from .llms.anthropic.common_utils import is_claude_tool_target
 from .llms.azure.audio_transcriptions import AzureAudioTranscription
 from .llms.azure.azure import AzureChatCompletion, check_dynamic_azure_params
 from .llms.azure.chat.o_series_handler import AzureOpenAIO1ChatCompletion
@@ -1135,17 +1137,6 @@ from litellm.responses.bridge_check import (  # noqa: E402  # re-export at the o
 _ANTHROPIC_ONLY_TOOL_KEYS: Final = frozenset({"input_examples", "eager_input_streaming"})
 
 
-def _is_claude_tool_target(custom_llm_provider: str | None, model: str) -> bool:
-    if custom_llm_provider == "anthropic":
-        return True
-    model_lower: Final = model.lower()
-    if custom_llm_provider == "bedrock":
-        return "claude" in model_lower or ("arn:" in model_lower and ":bedrock:" in model_lower)
-    if custom_llm_provider == "azure_ai" or custom_llm_provider == "vertex_ai":
-        return "claude" in model_lower
-    return False
-
-
 def _history_without_unanswered_tool_search_calls(
     messages: Sequence[object], custom_llm_provider: str | None, provider_config: BaseConfig
 ) -> tuple[object, ...] | None:
@@ -1282,6 +1273,16 @@ def _get_router_deployment_id(kwargs: dict) -> str | None:
     return None
 
 
+def _record_agentic_loop_params(
+    logging_obj: object, requested_model: str, custom_llm_provider: str, owns_agentic_loop: bool
+) -> None:
+    if not owns_agentic_loop or not isinstance(logging_obj, LiteLLMLoggingObj):
+        return
+    logging_obj.record_agentic_loop_params(
+        AgenticLoopParams(model=requested_model, custom_llm_provider=custom_llm_provider)
+    )
+
+
 def _register_custom_pricing_for_request(
     model: str,
     custom_llm_provider: str,
@@ -5701,6 +5702,7 @@ def completion(
             **{k: kwargs[k] for k in OPTIONAL_KWARGS_KEYS | {"provider_specific_header"} if k in kwargs},
             "extra_headers": extra_headers,
         }
+        requested_model: Final = model
         model, custom_llm_provider, dynamic_api_key, api_base = get_llm_provider(
             model=model,
             custom_llm_provider=custom_llm_provider,
@@ -5712,16 +5714,23 @@ def completion(
                 else None
             ),
         )
+        _record_agentic_loop_params(
+            logging_obj=logging,
+            requested_model=requested_model,
+            custom_llm_provider=custom_llm_provider,
+            owns_agentic_loop=not skip_responses_api_bridge,
+        )
 
+        deployment_model_info: Final = str_keyed_mapping(
+            cast(object, model_info)  # cast-ok: model_info is the untyped dict kwarg
+        )
         ## RESPONSES API BRIDGE LOGIC ## - check early and normalize model name
         responses_api_model_info, model = responses_api_bridge_check(
             model=model,
             custom_llm_provider=custom_llm_provider,
             web_search_options=web_search_options,
             api_base=api_base,
-            deployment_model_info=str_keyed_mapping(
-                cast(object, model_info)  # cast-ok: model_info is the untyped dict kwarg
-            ),
+            deployment_model_info=deployment_model_info,
         )
         routing_kwargs: Final = cast("Mapping[str, object]", kwargs)  # cast-ok: kwargs is the untyped request dict
         drop_unsupported_reasoning: Final = (
@@ -5752,7 +5761,7 @@ def completion(
             }
         )
 
-        if not _is_claude_tool_target(custom_llm_provider=custom_llm_provider, model=model):
+        if not is_claude_tool_target(custom_llm_provider=custom_llm_provider, model=model):
             tools = _drop_anthropic_only_tool_keys(tools=tools)
 
         if provider_specific_header is not None:
@@ -5818,7 +5827,15 @@ def completion(
             )
 
         if provider_config is not None:
-            translated_messages: Final = provider_config.translate_developer_role_to_system_role(messages=messages)
+            translated_messages: Final = list(
+                provider_config.translate_developer_role_to_system_role(
+                    messages=messages,
+                    model=model,
+                    custom_llm_provider=custom_llm_provider,
+                    api_base=api_base,
+                    model_info=deployment_model_info,
+                )
+            )
             history_for_route: Final = _history_without_unanswered_tool_search_calls(
                 translated_messages, custom_llm_provider, provider_config
             )

litellm/messages/dispatch.py

@@ -111,6 +111,7 @@ _FEATURES: Final[detectors.Detectors] = MappingProxyType(
         Feature.CACHE_CONTROL_INJECTION: detectors.cache_control_injection,
         Feature.STREAM_COST_INJECTION: route_detectors.stream_cost_injection,
         Feature.THINKING_SIGNATURE_RETRY: route_detectors.thinking_signature_retry,
+        Feature.TOOL_CUSTOM_FIELD: route_detectors.tool_custom_field,
     }
 )
 

litellm/provider_endpoints_support_backup.json

@@ -5327,6 +5327,32 @@
         "generateContent": false
       }
     },
+    "kyma": {
+      "display_name": "Kyma API (`kyma`)",
+      "url": "https://docs.litellm.ai/docs/providers/kyma",
+      "endpoints": {
+        "chat_completions": true,
+        "responses": true,
+        "text_completion": true,
+        "embeddings": false,
+        "image_generations": false,
+        "image_edits": false,
+        "audio_transcriptions": false,
+        "audio_speech": false,
+        "rerank": false,
+        "ocr": false,
+        "search": false,
+        "moderations": false,
+        "realtime": false,
+        "video_generations": false,
+        "batches": false,
+        "messages": true,
+        "interactions": true,
+        "systemone": false,
+        "a2a": true,
+        "generateContent": true
+      }
+    },
     "bigmodel": {
       "display_name": "BigModel (Zhipu AI) (`bigmodel`)",
       "url": "https://docs.bigmodel.cn/cn/guide/develop/responses/introduction",

litellm/proxy/_lazy_openapi_snapshot.json

@@ -7816,7 +7816,7 @@
     "paths": {
       "/claude-code/marketplace.json": {
         "get": {
-          "description": "Serve marketplace.json for Claude Code plugin discovery.\n\nThis endpoint is accessed by Claude Code CLI when users run:\n- claude plugin marketplace add <url>\n- claude plugin install <name>@<marketplace>\n\nWithout `key` the catalog holds the enabled (public) plugins. With `?key=sk-...`\nthe key is authenticated and the catalog also holds the disabled plugins granted\nto it through `object_permission.skills` or an access group on the key or its team.\n\nReturns:\n    Marketplace catalog with list of available plugins and their git sources.\n\nExample:\n    ```bash\n    claude plugin marketplace add http://localhost:4000/claude-code/marketplace.json\n    claude plugin marketplace add \"http://localhost:4000/claude-code/marketplace.json?key=sk-...\"\n    claude plugin install my-plugin@litellm\n    ```",
+          "description": "Serve marketplace.json for Claude Code plugin discovery.\n\nThis endpoint is accessed by Claude Code CLI when users run:\n- claude plugin marketplace add <url>\n- claude plugin install <name>@<marketplace>\n\nWithout a key the catalog holds the enabled (public) plugins. With `?key=sk-...`\nthe key is authenticated and the catalog also holds the disabled plugins granted\nto it through `object_permission.skills` or an access group on the key or its team.\n\nWith `general_settings.claude_code_marketplace_auth_required: true` the key must come\nin the `Authorization` or `x-litellm-api-key` header (Claude Code sends them from the\nmarketplace source's `headers`), `?key=` is ignored, and a caller with no valid key\ngets a 401 with `WWW-Authenticate: Bearer`.\n\nReturns:\n    Marketplace catalog with list of available plugins and their git sources.\n\nExample:\n    ```bash\n    claude plugin marketplace add http://localhost:4000/claude-code/marketplace.json\n    claude plugin marketplace add \"http://localhost:4000/claude-code/marketplace.json?key=sk-...\"\n    claude plugin install my-plugin@litellm\n    ```",
           "operationId": "get_marketplace_claude_code_marketplace_json_get",
           "parameters": [
             {
@@ -34706,7 +34706,7 @@
           ]
         },
         "get": {
-          "description": "Get a response by ID.\n\nSupports both:\n- Polling IDs (litellm_poll_*): Returns cumulative cached content from background responses\n- Provider response IDs: Passes through to provider API\n\nFollows the OpenAI Responses API spec: https://platform.openai.com/docs/api-reference/responses/get\n\n```bash\n# Get polling response\ncurl -X GET http://localhost:4000/v1/responses/litellm_poll_abc123     -H \"Authorization: Bearer $LITELLM_MASTER_KEY\"\n\n# Get provider response\ncurl -X GET http://localhost:4000/v1/responses/resp_abc123     -H \"Authorization: Bearer $LITELLM_MASTER_KEY\"\n```",
+          "description": "Get a response by ID.\n\nSupports both:\n- Polling IDs (litellm_poll_*): Returns cumulative cached content from background responses\n- Provider response IDs: Passes through to provider API\n\nFollows the OpenAI Responses API spec: https://platform.openai.com/docs/api-reference/responses/get\n\n```bash\n# Get polling response\ncurl -X GET http://localhost:4000/v1/responses/litellm_poll_abc123     -H \"Authorization: Bearer $LITELLM_MASTER_KEY\"\n\n# Get provider response\ncurl -X GET http://localhost:4000/v1/responses/resp_abc123     -H \"Authorization: Bearer $LITELLM_MASTER_KEY\"\n\n# Resume a background response's event stream after sequence number 7\ncurl -N \"http://localhost:4000/v1/responses/resp_abc123?stream=true&starting_after=7\"     -H \"Authorization: Bearer $LITELLM_MASTER_KEY\"\n```",
           "operationId": "get_response_openai_v1_responses__response_id__get",
           "parameters": [
             {
@@ -34717,6 +34717,46 @@
                 "title": "Response Id",
                 "type": "string"
               }
+            },
+            {
+              "description": "Replay the response's events as server-sent events (OpenAI, Azure, and Azure AI Foundry)",
+              "in": "query",
+              "name": "stream",
+              "required": false,
+              "schema": {
+                "type": "boolean"
+              }
+            },
+            {
+              "description": "With stream=true, replay only the events after this sequence number",
+              "in": "query",
+              "name": "starting_after",
+              "required": false,
+              "schema": {
+                "minimum": 0,
+                "type": "integer"
+              }
+            },
+            {
+              "description": "Additional output data to include, forwarded to the provider",
+              "in": "query",
+              "name": "include",
+              "required": false,
+              "schema": {
+                "items": {
+                  "type": "string"
+                },
+                "type": "array"
+              }
+            },
+            {
+              "description": "With stream=true, whether delta events carry obfuscation padding",
+              "in": "query",
+              "name": "include_obfuscation",
+              "required": false,
+              "schema": {
+                "type": "boolean"
+              }
             }
           ],
           "responses": {
@@ -34726,6 +34766,12 @@
                   "schema": {
                     "$ref": "#/components/schemas/ResponsesAPIResponse"
                   }
+                },
+                "text/event-stream": {
+                  "schema": {
+                    "description": "Server sent events when stream=true",
+                    "type": "string"
+                  }
                 }
               },
               "description": "Successful Response"

litellm/proxy/_types.py

@@ -2833,6 +2833,10 @@ class ConfigGeneralSettings(LiteLLMPydanticObjectBase):
         None,
         description="If True, stores request messages and responses in spend logs. Default is False.",
     )
+    claude_code_marketplace_auth_required: bool = Field(
+        False,
+        description="If True, GET /claude-code/marketplace.json requires a LiteLLM key in the Authorization or x-litellm-api-key header (?key= is not accepted) and answers 401 with WWW-Authenticate: Bearer without one, and GET /public/skill_hub returns an empty list. Default is False, which serves the enabled plugins to anyone.",
+    )
     disable_requester_ip_address_logging: bool | None = Field(
         None,
         description="If True, new spend log rows store requester_ip_address as null in the column, the metadata and the stored request. The stored request also drops client-IP forwarding headers such as x-forwarded-for, and IP addresses in its other header values and in stored error messages are masked. Guardrails, metrics and callbacks that read the standard logging payload still receive the IP. Default is False.",
@@ -3744,6 +3748,7 @@ class SpendLogsMetadata(TypedDict):
     internal_call_origin: InternalCallOrigin | None
     billed_retry_attempt: ReadOnly[NotRequired[bool | None]]
     litellm_roi_estimator: ReadOnly[NotRequired[bool | None]]
+    is_silent_experiment: ReadOnly[NotRequired[bool | None]]
     guardrail_information: list[StandardLoggingGuardrailInformation] | None
     eval_information: Any | None
     status: StandardLoggingPayloadStatus

litellm/proxy/anthropic_endpoints/claude_code_endpoints/claude_code_marketplace.py

@@ -7,7 +7,8 @@ Actual plugin files are hosted on GitHub/GitLab/Bitbucket or as a zip archive on
 any HTTPS host (S3, Artifactory, a static file server).
 
 Endpoints:
-/claude-code/marketplace.json  - GET  - List plugins for Claude Code discovery (unauthenticated; `?key=` adds the key's granted skills)
+/claude-code/marketplace.json  - GET  - List plugins for Claude Code discovery (anonymous unless
+                                 general_settings.claude_code_marketplace_auth_required; a key adds its granted skills)
 /claude-code/plugins           - POST - Register a new plugin (create-only, proxy admin only)
 /claude-code/plugins           - GET  - List plugins visible to the key (enabled, plus granted disabled ones)
 /claude-code/plugins/{name}    - GET  - Get plugin details (403 on a disabled plugin the key is not granted)
@@ -31,14 +32,17 @@ from pydantic import TypeAdapter
 from typing_extensions import ReadOnly
 
 from litellm._logging import verbose_proxy_logger
-from litellm.proxy._types import CommonProxyErrors, ProxyException, UserAPIKeyAuth
+from litellm.proxy._types import CommonProxyErrors, ProxyErrorTypes, ProxyException, UserAPIKeyAuth
 from litellm.proxy.anthropic_endpoints.claude_code_endpoints.claude_code_skill_access import (
     SkillVisibility,
     resolve_skill_visibility,
 )
+from litellm.proxy.auth.auth_exception_handler import DB_UNAVAILABLE_FALLBACK_USER_ID
 from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
+from litellm.proxy.common_utils.claude_code_marketplace_auth import marketplace_auth_required
 from litellm.proxy.common_utils.resource_ownership import is_proxy_admin
 from litellm.repositories.table_repositories import ClaudeCodePluginRepository
+from litellm.types.proxy.auth.special_headers import SpecialHeaders
 from litellm.types.proxy.claude_code_endpoints import (
     InstallationPreference,
     ListPluginsResponse,
@@ -109,6 +113,41 @@ async def get_prisma_client() -> object:
 _get_prisma_client: Final = get_prisma_client
 
 
+async def _header_key_caller(request: Request) -> UserAPIKeyAuth:
+    try:
+        caller: Final = await user_api_key_auth(
+            request=request,
+            api_key=request.headers.get(SpecialHeaders.openai_authorization.value) or "",
+            custom_litellm_key_header=request.headers.get(SpecialHeaders.custom_litellm_api_key.value),
+        )
+    except ProxyException as e:
+        if e.code != "401":
+            raise
+        raise ProxyException(
+            message=e.message,
+            type=e.type,
+            param=e.param,
+            code=401,
+            headers={**e.headers, "WWW-Authenticate": "Bearer"},
+        ) from e
+    if caller.user_id == DB_UNAVAILABLE_FALLBACK_USER_ID:
+        raise ProxyException(
+            message="The LiteLLM key could not be verified because the database is unavailable",
+            type=ProxyErrorTypes.no_db_connection,
+            param=None,
+            code=503,
+        )
+    return caller
+
+
+async def _marketplace_caller(request: Request, key: str | None) -> UserAPIKeyAuth | None:
+    if marketplace_auth_required():
+        return await _header_key_caller(request)
+    if key:
+        return await user_api_key_auth(request=request, api_key=f"Bearer {key}")
+    return None
+
+
 @router.get(
     "/claude-code/marketplace.json",
     tags=["Claude Code Marketplace"],
@@ -121,10 +160,15 @@ async def get_marketplace(request: Request, key: str | None = None):
     - claude plugin marketplace add <url>
     - claude plugin install <name>@<marketplace>
 
-    Without `key` the catalog holds the enabled (public) plugins. With `?key=sk-...`
+    Without a key the catalog holds the enabled (public) plugins. With `?key=sk-...`
     the key is authenticated and the catalog also holds the disabled plugins granted
     to it through `object_permission.skills` or an access group on the key or its team.
 
+    With `general_settings.claude_code_marketplace_auth_required: true` the key must come
+    in the `Authorization` or `x-litellm-api-key` header (Claude Code sends them from the
+    marketplace source's `headers`), `?key=` is ignored, and a caller with no valid key
+    gets a 401 with `WWW-Authenticate: Bearer`.
+
     Returns:
         Marketplace catalog with list of available plugins and their git sources.
 
@@ -137,10 +181,7 @@ async def get_marketplace(request: Request, key: str | None = None):
     """
     try:
         prisma_client: Final = await get_prisma_client()
-
-        caller: Final[UserAPIKeyAuth | None] = (
-            await user_api_key_auth(request=request, api_key=f"Bearer {key}") if key else None
-        )
+        caller: Final = await _marketplace_caller(request, key)
         visibility: Final[SkillVisibility] = await resolve_skill_visibility(caller)
         plugins: Final[Sequence[_PluginRecord]] = await ClaudeCodePluginRepository(prisma_client).table.find_many(
             where=visibility.where()
@@ -201,7 +242,7 @@ async def get_marketplace(request: Request, key: str | None = None):
         verbose_proxy_logger.exception("Error generating marketplace: %s", e)
         raise HTTPException(
             status_code=500,
-            detail={"error": f"Failed to generate marketplace: {e}"},
+            detail={"error": "Failed to generate marketplace"},
         )
 
 

litellm/proxy/anthropic_endpoints/endpoints.py

@@ -2,7 +2,7 @@
 Unified /v1/messages endpoint - (Anthropic Spec)
 """
 
-from collections.abc import Mapping
+from collections.abc import AsyncGenerator, Mapping
 from types import MappingProxyType
 from typing import Final
 
@@ -19,12 +19,10 @@ from litellm.anthropic_interface.exceptions import (
 )
 from litellm.integrations.custom_guardrail import ModifyResponseException
 from litellm.litellm_core_utils.tool_fields import first_validation_error
+from litellm.llms.anthropic.chat.guardrail_translation.handler import AnthropicMessagesHandler
 from litellm.llms.anthropic.messages.context_management import (
     AnthropicContextManagementError,
 )
-from litellm.llms.base_llm.guardrail_translation.utils import (
-    blocked_response_usage as _blocked_response_usage,
-)
 from litellm.proxy._types import *
 from litellm.proxy.auth.user_api_key_auth import user_api_key_auth
 from litellm.proxy.common_request_processing import (
@@ -44,6 +42,7 @@ from litellm.proxy.common_utils.http_parsing_utils import (  # noqa: F401  # leg
 from litellm.proxy.common_utils.openai_error_payload import (
     LITELLM_CALL_ID_HEADER,
     error_status_code,
+    litellm_call_id_headers,
     openai_error_param,
     openai_error_type,
     with_litellm_call_id,
@@ -58,6 +57,8 @@ from litellm.router import Router
 
 router: Final = APIRouter()
 
+_NO_HEADERS: Final[Mapping[str, str]] = MappingProxyType({})
+
 
 def _with_provider_specific_fields(exc: ProxyException, detail: AnthropicErrorDetail) -> AnthropicErrorDetail:
     if not exc.provider_specific_fields:
@@ -229,44 +230,22 @@ async def anthropic_response(
             request_data=_data,
         )
 
-        # Create Anthropic-formatted response with violation message
-        import uuid
+        block_headers: Final = litellm_call_id_headers(base_llm_response_processor.litellm_call_id) or _NO_HEADERS
+        block_handler: Final = AnthropicMessagesHandler()
 
-        from litellm.types.utils import AnthropicMessagesResponse
+        if data.get("stream") is True:
+            block_chunks: Final = block_handler.build_block_sse_chunks(e)
 
-        # Report the blocked LLM response's real token usage (carried on the
-        # exception) instead of discarding it; zero for pre-call blocks.
-        _usage: Final = _blocked_response_usage(e.original_response)
-
-        _anthropic_response: Final = AnthropicMessagesResponse(
-            id=f"msg_{uuid.uuid4()}",
-            type="message",
-            role="assistant",
-            content=[{"type": "text", "text": e.message}],
-            model=e.model,
-            stop_reason="end_turn",
-            usage=_usage,
-        )
-
-        if data.get("stream", None) is not None and data["stream"] is True:
-            # For streaming, use the standard SSE data generator
-            async def _passthrough_stream_generator():
-                yield _anthropic_response
-
-            selected_data_generator: Final = ProxyBaseLLMRequestProcessing.async_sse_data_generator(
-                response=_passthrough_stream_generator(),
-                user_api_key_dict=user_api_key_dict,
-                request_data=_data,
-                proxy_logging_obj=proxy_logging_obj,
-            )
+            async def _blocked_stream() -> AsyncGenerator[str, None]:
+                for chunk in block_chunks:
+                    yield chunk.decode()
 
             return await create_response(
-                generator=selected_data_generator,
-                media_type="text/event-stream",
-                headers={},
+                generator=_blocked_stream(), media_type="text/event-stream", headers=block_headers
             )
 
-        return _anthropic_response
+        fastapi_response.headers.update(block_headers)
+        return block_handler.build_block_response(e)
     except AnthropicContextManagementError as e:
         if e.status_code >= 500:
             # Server-side polyfill failures hit the failure hook for spend/alert

litellm/proxy/anthropic_endpoints/gateway_endpoints.py

@@ -45,7 +45,11 @@ from litellm.proxy.common_utils.http_parsing_utils import (  # noqa: F401  # leg
     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.proxy.management_endpoints.ui_sso import (
+    CLI_SSO_BROWSER_TEAM_CHOICE,
+    CLI_SSO_CHOSEN_TEAM_ID,
+    CliSsoTeamDetail,
+)
 from litellm.proxy.mcp_server.gateway_dcr_flow import (
     PROXY_API_AUDIENCE,
     MintedProxyCredential,
@@ -235,6 +239,7 @@ async def device_authorization(request: Request) -> JSONResponse:
         "sso_complete": False,
         "user_code_verified": False,
         "session_data": None,
+        CLI_SSO_BROWSER_TEAM_CHOICE: True,
     }
     _set_cli_sso_flow(login_id=login_id, cache=cli_sso_session_cache, flow=flow)
 
@@ -267,7 +272,15 @@ def _validate_login(flow: Mapping[str, object]) -> _GatewayLogin | _OAuthError:
             status_code=400, error="invalid_grant", description="The login session is malformed; sign in again"
         )
 
-    team_id: Final = session_data.teams[0] if session_data.teams else None
+    chosen: Final = flow.get(CLI_SSO_CHOSEN_TEAM_ID)
+    chosen_team_id: Final = chosen if isinstance(chosen, str) else None
+    if chosen_team_id is not None and chosen_team_id not in session_data.teams:
+        return _OAuthError(
+            status_code=400,
+            error="invalid_grant",
+            description="The team picked at sign-in is no longer one of your teams; sign in again",
+        )
+    team_id: Final = chosen_team_id or next(iter(session_data.teams), None)
     selected_team: Final = selected_cli_sso_team_detail(team_details=session_data.team_details, team_id=team_id)
     if selected_team is None:
         return _OAuthError(

litellm/proxy/common_request_processing.py

@@ -3206,7 +3206,7 @@ class ProxyBaseLLMRequestProcessing:
                         request_data=self.data,
                         request=request,
                     )
-                    if route_type == "aresponses":
+                    if route_type in ("aresponses", "aget_responses"):
                         selected_data_generator = (
                             ProxyBaseLLMRequestProcessing._wrap_responses_stream_for_container_ownership(
                                 original_stream_response=response,

litellm/proxy/common_utils/claude_code_marketplace_auth.py

@@ -0,0 +1,17 @@
+from typing import Final
+
+from pydantic import TypeAdapter, ValidationError
+
+_AUTH_REQUIRED_SETTING: Final[TypeAdapter[bool | None]] = TypeAdapter(bool | None)
+
+
+def marketplace_auth_required() -> bool:
+    from litellm.proxy.proxy_server import general_settings_view
+
+    try:
+        return (
+            _AUTH_REQUIRED_SETTING.validate_python(general_settings_view().get("claude_code_marketplace_auth_required"))
+            is True
+        )
+    except ValidationError:
+        return True

litellm/proxy/common_utils/html_forms/native_client_consent.py

@@ -62,7 +62,7 @@ button {{ flex: 1; padding: 10px; border-radius: 6px; font-size: 15px; cursor: p
 <p>Approving issues it a personal credential that expires within {CLI_JWT_EXPIRATION_HOURS} hours. <code>lite logout</code> stops it from being renewed. Only approve if you started this sign-in yourself.</p>
 <form method="post" action="{escape(complete_url)}">
 <input type="hidden" name="flow" value="{escape(flow_handle)}">
-{_team_field(teams)}
+{team_select_field(teams)}
 <div class="actions">
 <button type="submit" name="decision" value="deny" class="deny">Deny</button>
 <button type="submit" name="decision" value="approve" class="approve">Approve</button>
@@ -74,7 +74,7 @@ button {{ flex: 1; padding: 10px; border-radius: 6px; font-size: 15px; cursor: p
 """
 
 
-def _team_field(teams: Sequence[tuple[str, str]]) -> str:
+def team_select_field(teams: Sequence[tuple[str, str]], *, require_choice: bool = False) -> str:
     if not teams:
         return ""
     if len(teams) == 1:
@@ -83,9 +83,12 @@ def _team_field(teams: Sequence[tuple[str, str]]) -> str:
             f'<input type="hidden" name="team_id" value="{escape(team_id)}">'
             f"<p>Requests are attributed to team <strong>{escape(team_label)}</strong>.</p>"
         )
+    placeholder: Final = '<option value="" disabled selected>Choose a team</option>' if require_choice else ""
     options: Final = "".join(
         f'<option value="{escape(team_id)}">{escape(team_label)}</option>' for team_id, team_label in teams
     )
+    required: Final = " required" if require_choice else ""
     return (
-        f'<label for="team_id">Attribute requests to team</label><select id="team_id" name="team_id">{options}</select>'
+        '<label for="team_id">Attribute requests to team</label>'
+        f'<select id="team_id" name="team_id"{required}>{placeholder}{options}</select>'
     )

litellm/proxy/db/check_migration.py

@@ -4,7 +4,7 @@ import os
 import subprocess
 from typing import Final
 
-from litellm._logging import print_startup_message, verbose_logger
+from litellm._logging import print_startup_message, print_startup_warning, verbose_logger
 
 
 def extract_sql_commands(diff_output: str) -> list[str]:
@@ -63,7 +63,7 @@ def check_prisma_schema_diff_helper(db_url: str) -> tuple[bool, list[str]]:
             run_prisma,
         )
     except ImportError as e:
-        print_startup_message(
+        print_startup_warning(
             f"Skipping the migration diff: litellm-proxy-extras has no Prisma runner. Error: {e}",
         )
         return False, []
@@ -96,14 +96,14 @@ def check_prisma_schema_diff_helper(db_url: str) -> tuple[bool, list[str]]:
         else:
             return False, []
     except subprocess.TimeoutExpired:
-        print_startup_message(
+        print_startup_warning(
             f"Timed out after {timeout}s generating the migration diff. "
             f"Raise {PRISMA_COMMAND_TIMEOUT_ENV_VAR} if this database needs longer.",
         )
         return False, []
     except subprocess.CalledProcessError as e:
         error_message: Final = f"Failed to generate migration diff. Error: {e.stderr}"
-        print_startup_message(error_message)
+        print_startup_warning(error_message)
         return False, []
 
 

litellm/proxy/db/db_transaction_queue/spend_log_cleanup.py

@@ -421,7 +421,10 @@ class SpendLogCleanup:
         general_settings=None,
         redis_cache: RedisCache | None = None,
         partition_manager: SpendLogsPartitionManager | None = None,
+        *,
+        clock: Callable[[], float] = time.monotonic,
     ):
+        self._clock = clock
         self.retention_seconds: int | None = None
         self.partition_manager = partition_manager or SpendLogsPartitionManager()
         from litellm.proxy.proxy_server import general_settings as default_settings
@@ -586,7 +589,7 @@ class SpendLogCleanup:
 
     async def _probe_storage(self, prisma_client: PrismaClient, deadline: float) -> StorageProbe | None:
         """Read what the spend-log tables weigh from the catalog, or None when the probe fails or has no budget."""
-        if time.monotonic() >= deadline:
+        if self._clock() >= deadline:
             return None
         timeout_ms: Final = self._timeout_ms(deadline)
         try:
@@ -624,7 +627,7 @@ class SpendLogCleanup:
             stale.allocated_bytes,
             _ANALYZE_SQL,
         )
-        if time.monotonic() >= deadline:
+        if self._clock() >= deadline:
             return None
         timeout_ms: Final = self._timeout_ms(deadline)
         try:
@@ -653,7 +656,7 @@ class SpendLogCleanup:
 
     async def _load_size_state(self, prisma_client: PrismaClient, deadline: float) -> SizeCleanupState | None:
         """The reclaimable count saved by the last check, or None on the first check or when it cannot be read."""
-        if time.monotonic() >= deadline:
+        if self._clock() >= deadline:
             return None
         timeout_ms: Final = self._timeout_ms(deadline)
         try:
@@ -686,7 +689,7 @@ class SpendLogCleanup:
         the tool-index rows that expire with their spend logs. When fewer rows than that remain, the step is
         everything up to now.
         """
-        if time.monotonic() >= deadline:
+        if self._clock() >= deadline:
             return None
         timeout_ms: Final = self._timeout_ms(deadline)
         async with db_span("size_step_cutoff", "LiteLLM_SpendLogs"), bounded_tx(prisma_client, timeout_ms) as tx:
@@ -718,11 +721,11 @@ class SpendLogCleanup:
         Interpolating this into SQL is safe by construction: an int cannot carry
         SQL, and SET does not accept a bind parameter.
         """
-        remaining_ms: Final = int((deadline - time.monotonic()) * 1000)
+        remaining_ms: Final = int((deadline - self._clock()) * 1000)
         return max(1, min(int(self.batch_timeout_seconds * 1000), remaining_ms))
 
     @staticmethod
-    def _group_deadline(overall_deadline: float, groups_remaining: int) -> float:
+    def _group_deadline(overall_deadline: float, groups_remaining: int, now: float) -> float:
         """
         Give each pending cleanup group an equal share of the time left.
 
@@ -731,10 +734,9 @@ class SpendLogCleanup:
         """
         if groups_remaining == 1:
             return overall_deadline
-        current_time: Final = time.monotonic()
-        if current_time >= overall_deadline:
+        if now >= overall_deadline:
             return overall_deadline
-        return current_time + (overall_deadline - current_time) / groups_remaining
+        return now + (overall_deadline - now) / groups_remaining
 
     def _remaining_timeout_ms(self, deadline: float) -> RemainingTimeoutMs:
         """
@@ -748,7 +750,7 @@ class SpendLogCleanup:
         """
 
         def remaining() -> int | None:
-            return None if time.monotonic() >= deadline else self._timeout_ms(deadline)
+            return None if self._clock() >= deadline else self._timeout_ms(deadline)
 
         return remaining
 
@@ -841,7 +843,7 @@ class SpendLogCleanup:
         run_count = 0
         consecutive_failures = 0
         while True:
-            if time.monotonic() >= deadline:
+            if self._clock() >= deadline:
                 verbose_proxy_logger.info(
                     "Run budget exhausted during %s cleanup after %d rows; the next run resumes from here",
                     table_name,
@@ -858,13 +860,13 @@ class SpendLogCleanup:
                     prisma_client, cutoff_date, table_name, time_column, total_deleted, "batch_cap_reached", deadline
                 )
             # Find rows and delete them in one go without fetching to application
-            batch_started_at = time.monotonic()
+            batch_started_at = self._clock()
             try:
                 batch_result = await self._execute_delete_batch(
                     prisma_client, delete_sql, cutoff_date, table_name, deadline
                 )
             except Exception as batch_exc:
-                if time.monotonic() >= deadline:
+                if self._clock() >= deadline:
                     # The statement timeout was clamped to the budget that was
                     # left, so this batch was cancelled by the deadline itself.
                     # That is the bound working, not a database fault, and
@@ -920,7 +922,7 @@ class SpendLogCleanup:
 
             consecutive_failures = 0
             deleted_count = batch_result
-            SpendLogCleanupMetrics.record_batch(table_name, deleted_count, time.monotonic() - batch_started_at)
+            SpendLogCleanupMetrics.record_batch(table_name, deleted_count, self._clock() - batch_started_at)
             verbose_proxy_logger.info("Deleted %s %s rows in this batch", deleted_count, table_name)
 
             if deleted_count == 0:
@@ -957,7 +959,7 @@ class SpendLogCleanup:
         operator the backlog was not drained; the gauge simply keeps its value
         from the last run that finished inside its budget.
         """
-        if time.monotonic() >= deadline:
+        if self._clock() >= deadline:
             return TableCleanupResult(table_name=table_name, rows_deleted=rows_deleted, stop_reason=stop_reason)
         remaining: Final = await self._count_remaining(prisma_client, cutoff_date, table_name, time_column, deadline)
         if remaining is not None:
@@ -1070,7 +1072,7 @@ class SpendLogCleanup:
         # behind any long-running reader for as long as that reader lives, which
         # is the one way this job could still outlast its budget without bound.
         remaining_timeout_ms: Final = self._remaining_timeout_ms(deadline)
-        if time.monotonic() >= deadline:
+        if self._clock() >= deadline:
             verbose_proxy_logger.info("Run budget already spent, skipping partition maintenance this run")
         elif self.general_settings.get(
             "use_spend_logs_partitioning", False
@@ -1082,7 +1084,7 @@ class SpendLogCleanup:
             verbose_proxy_logger.info("Dropped %d expired spend-log partitions: %s", len(dropped), dropped)
 
         logs_result: Final = await self._delete_old_logs(
-            prisma_client, cutoff_date, self._group_deadline(deadline, groups_remaining=2)
+            prisma_client, cutoff_date, self._group_deadline(deadline, groups_remaining=2, now=self._clock())
         )
         verbose_proxy_logger.info("Deleted %s logs", logs_result.rows_deleted)
 
@@ -1210,7 +1212,7 @@ class SpendLogCleanup:
         self, prisma_client: PrismaClient, deadline: float, still_over_target: Callable[[], Awaitable[bool]]
     ) -> tuple[str, ...]:
         remaining_timeout_ms: Final = self._remaining_timeout_ms(deadline)
-        if time.monotonic() >= deadline or not self._setting("use_spend_logs_partitioning"):
+        if self._clock() >= deadline or not self._setting("use_spend_logs_partitioning"):
             return ()
         if not await self.partition_manager.is_partitioned(prisma_client, remaining_timeout_ms):
             return ()
@@ -1257,7 +1259,7 @@ class SpendLogCleanup:
         needed: Final = rows_still_needed(state.logs_deleted + state.index_deleted)
         if needed <= 0:
             return state.stopped("target_reached")
-        if time.monotonic() >= deadline:
+        if self._clock() >= deadline:
             return state.stopped("budget_exhausted")
         if state.batches_left <= 0:
             return state.stopped("batch_cap_reached")
@@ -1318,11 +1320,11 @@ class SpendLogCleanup:
             except Exception:  # noqa: BLE001  # retained observations are retried by the next cleanup job
                 verbose_proxy_logger.warning("Auto-router baseline retention remains pending")
         sessions_result: Final = await self._delete_old_autorouter_session_rows(
-            prisma_client, session_cutoff, self._group_deadline(deadline, 3)
+            prisma_client, session_cutoff, self._group_deadline(deadline, 3, self._clock())
         )
         verbose_proxy_logger.info("Deleted %s expired auto-router session rollup rows", sessions_result.rows_deleted)
         user_sessions_result: Final = await self._delete_old_autorouter_user_session_rows(
-            prisma_client, session_cutoff, self._group_deadline(deadline, 2)
+            prisma_client, session_cutoff, self._group_deadline(deadline, 2, self._clock())
         )
         verbose_proxy_logger.info(
             "Deleted %s expired auto-router user session rollup rows", user_sessions_result.rows_deleted
@@ -1443,7 +1445,7 @@ class SpendLogCleanup:
 
     async def _run_passes(self, prisma_client: PrismaClient, age_rules: bool, size_rule: bool) -> None:
         lock_acquired = False
-        run_started_at: Final = time.monotonic()
+        run_started_at: Final = self._clock()
         progress: Final = _RunProgress()
         progress_token: Final = _run_progress.set(progress)
         try:
@@ -1493,7 +1495,7 @@ class SpendLogCleanup:
                     SpendLogCleanupMetrics.record_run("skipped_locked")
                     return
 
-            deadline: Final = time.monotonic() + self.run_budget_seconds
+            deadline: Final = self._clock() + self.run_budget_seconds
             size_group: Final = int(active_size_rule is not None)
             configured_group_count: Final = (
                 int(delete_spend_logs and self.retention_seconds is not None)
@@ -1506,7 +1508,7 @@ class SpendLogCleanup:
             spend_log_results: Final = (
                 await self._clean_spend_log_tables(
                     prisma_client,
-                    self._group_deadline(deadline, configured_group_count),
+                    self._group_deadline(deadline, configured_group_count, self._clock()),
                 )
                 if delete_spend_logs and self.retention_seconds is not None
                 else ()
@@ -1526,7 +1528,7 @@ class SpendLogCleanup:
                 await self._clean_session_rollup(
                     prisma_client,
                     autorouter_retention_seconds,
-                    self._group_deadline(deadline, remaining_groups_after_spend_logs),
+                    self._group_deadline(deadline, remaining_groups_after_spend_logs, self._clock()),
                 )
                 if autorouter_retention_seconds is not None
                 else ()
@@ -1535,7 +1537,7 @@ class SpendLogCleanup:
                 await self._clean_health_checks(
                     prisma_client,
                     health_check_retention_seconds,
-                    self._group_deadline(deadline, remaining_groups_after_sessions),
+                    self._group_deadline(deadline, remaining_groups_after_sessions, self._clock()),
                 )
                 if health_check_retention_seconds is not None
                 else ()
@@ -1544,7 +1546,7 @@ class SpendLogCleanup:
                 await self._clean_daily_tag_spend(
                     prisma_client,
                     daily_tag_spend_retention_seconds,
-                    self._group_deadline(deadline, 1 + size_group),
+                    self._group_deadline(deadline, 1 + size_group, self._clock()),
                 )
                 if daily_tag_spend_retention_seconds is not None
                 else ()
@@ -1561,12 +1563,12 @@ class SpendLogCleanup:
             )
             outcome: Final = self._run_outcome(results)
             SpendLogCleanupMetrics.record_run(outcome)
-            self._log_run_summary(outcome, results, time.monotonic() - run_started_at, size_report)
+            self._log_run_summary(outcome, results, self._clock() - run_started_at, size_report)
 
         except asyncio.CancelledError:
             verbose_proxy_logger.error(
                 "Spend log cleanup cancelled after %.2fs (rows_deleted=%d, batches=%d); the next run resumes from here",
-                time.monotonic() - run_started_at,
+                self._clock() - run_started_at,
                 progress.rows_deleted,
                 progress.batches,
             )

litellm/proxy/discovery_endpoints/ui_discovery_endpoints.py

@@ -4,6 +4,7 @@ from typing import Final
 
 from fastapi import APIRouter
 
+from litellm.proxy.common_utils.claude_code_marketplace_auth import marketplace_auth_required
 from litellm.proxy.common_utils.html_forms.default_credentials_hint import should_hide_default_credentials_hint
 from litellm.proxy.mcp_server.stdio_gate import is_mcp_stdio_enabled
 from litellm.types.proxy.discovery_endpoints.ui_discovery_endpoints import (
@@ -51,4 +52,5 @@ async def get_ui_config():
         workers=proxy_config.worker_registry if is_control_plane else [],
         mcp_stdio_enabled=is_mcp_stdio_enabled(),
         use_si_terminology=use_si_terminology,
+        claude_code_marketplace_auth_required=marketplace_auth_required(),
     )

litellm/proxy/hooks/responses_id_security.py

@@ -31,7 +31,7 @@ if TYPE_CHECKING:
 
 
 _RESPONSES_API_PROVIDER_PREFIX: Final = "/openai"
-_RESPONSES_API_CREATE_ROUTES: Final = frozenset({"/v1/responses", "/responses"})
+_RESPONSES_API_ROUTES: Final = frozenset({"/v1/responses", "/responses"})
 _ADDRESSED_RESPONSE_ID_FIELD_BY_CALL_TYPE: Final[Mapping[str, str]] = MappingProxyType(
     {
         "aresponses": "previous_response_id",
@@ -104,7 +104,7 @@ def _rewrite_advertised_id(
     return event
 
 
-def _is_responses_api_create_route(request_route: str | None) -> bool:
+def _streams_response_ids(request_route: str | None) -> bool:
     if request_route is None:
         return False
     canonical: Final = (
@@ -112,7 +112,8 @@ def _is_responses_api_create_route(request_route: str | None) -> bool:
         if request_route.startswith(_RESPONSES_API_PROVIDER_PREFIX + "/")
         else request_route
     )
-    return canonical in _RESPONSES_API_CREATE_ROUTES
+    parent, _, response_id = canonical.rpartition("/")
+    return canonical in _RESPONSES_API_ROUTES or (parent in _RESPONSES_API_ROUTES and response_id != "")
 
 
 class ResponsesIDSecurity(CustomLogger):
@@ -339,7 +340,7 @@ class ResponsesIDSecurity(CustomLogger):
         async for chunk in response:
             if (
                 isinstance(chunk, BaseLiteLLMOpenAIResponseObject)
-                and _is_responses_api_create_route(user_api_key_dict.request_route)
+                and _streams_response_ids(user_api_key_dict.request_route)
                 and not general_settings.get("disable_responses_id_security", False)
             ):
                 chunk = self._encrypt_response_id(chunk, user_api_key_dict, request_encryption_cache)