chore(release): backport Fable 5, batch-file auth, CrowdStrike AIDR, Mantle Responses SigV4, and NetApp streaming-cost fix to stable/1.88.x and cut 1.88.2

litellm/__init__.py

@@ -1727,6 +1727,9 @@ def set_global_gitlab_config(config: Dict[str, Any]) -> None:
     from .llms.openrouter.responses.transformation import (
         OpenRouterResponsesAPIConfig as OpenRouterResponsesAPIConfig,
     )
+    from .llms.bedrock_mantle.responses.transformation import (
+        BedrockMantleResponsesAPIConfig as BedrockMantleResponsesAPIConfig,
+    )
     from .llms.gemini.interactions.transformation import (
         GoogleAIStudioInteractionsConfig as GoogleAIStudioInteractionsConfig,
     )

litellm/_lazy_imports_registry.py

@@ -237,6 +237,7 @@
     "PerplexityResponsesConfig",
     "DatabricksResponsesAPIConfig",
     "OpenRouterResponsesAPIConfig",
+    "BedrockMantleResponsesAPIConfig",
     "GoogleAIStudioInteractionsConfig",
     "OpenAIOSeriesConfig",
     "AnthropicSkillsConfig",
@@ -956,6 +957,10 @@
         ".llms.openrouter.responses.transformation",
         "OpenRouterResponsesAPIConfig",
     ),
+    "BedrockMantleResponsesAPIConfig": (
+        ".llms.bedrock_mantle.responses.transformation",
+        "BedrockMantleResponsesAPIConfig",
+    ),
     "GoogleAIStudioInteractionsConfig": (
         ".llms.gemini.interactions.transformation",
         "GoogleAIStudioInteractionsConfig",

litellm/constants.py

@@ -1147,6 +1147,7 @@
     "openai.gpt-oss-120b-1:0",
     "anthropic.claude-haiku-4-5-20251001-v1:0",
     "anthropic.claude-sonnet-4-5-20250929-v1:0",
+    "anthropic.claude-fable-5",
     "anthropic.claude-opus-4-8",
     "anthropic.claude-opus-4-7",
     "anthropic.claude-opus-4-6-v1:0",

litellm/litellm_core_utils/llm_cost_calc/tool_call_cost_tracking.py

@@ -6,6 +6,7 @@
 
 import litellm
 from litellm.constants import OPENAI_FILE_SEARCH_COST_PER_1K_CALLS
+from litellm.litellm_core_utils.llm_cost_calc.utils import _get_web_search_requests
 from litellm.types.llms.openai import (
     FileSearchTool,
     ResponsesAPIResponse,
@@ -339,8 +340,7 @@ def response_object_includes_web_search_call(
                 # and _handle_web_search_cost() is never called.
                 if (
                     hasattr(usage, "server_tool_use")
-                    and usage.server_tool_use is not None
-                    and usage.server_tool_use.web_search_requests is not None
+                    and _get_web_search_requests(usage.server_tool_use) is not None
                 ):
                     return True
             return False
@@ -352,8 +352,7 @@ def response_object_includes_web_search_call(
         elif usage is not None:
             if (
                 hasattr(usage, "server_tool_use")
-                and usage.server_tool_use is not None
-                and usage.server_tool_use.web_search_requests is not None
+                and _get_web_search_requests(usage.server_tool_use) is not None
             ):
                 return True
             elif (

litellm/litellm_core_utils/llm_cost_calc/utils.py

@@ -1,7 +1,7 @@
 # What is this?
 ## Helper utilities for cost_per_token()
 
-from typing import Literal, Optional, Tuple, TypedDict, cast
+from typing import Any, Literal, Optional, Tuple, TypedDict, cast
 
 import litellm
 from litellm._logging import verbose_logger
@@ -34,6 +34,34 @@
 _VALID_DATA_RESIDENCIES = frozenset(r.value for r in DataResidency)
 
 
+def _get_token_detail_value(details: object, key: str) -> Optional[int]:
+    if isinstance(details, dict):
+        value = details.get(key)
+    else:
+        value = getattr(details, key, None)
+    return value if isinstance(value, int) else None
+
+
+def _get_web_search_requests(server_tool_use: Any) -> Optional[int]:
+    """
+    Tolerantly read ``web_search_requests`` from a ``server_tool_use`` value
+    that may be ``None``, a ``dict``, a ``ServerToolUse`` pydantic instance,
+    or any other object supporting attribute access.
+
+    Returns ``None`` when the value cannot be resolved — callers can
+    distinguish "absent" from "zero" using ``is None``.
+
+    See https://github.com/BerriAI/litellm/issues/26153 — ``stream_chunk_builder``
+    historically left this as a plain ``dict``, which broke direct attribute
+    access in cost calculation.
+    """
+    if server_tool_use is None:
+        return None
+    if isinstance(server_tool_use, dict):
+        return server_tool_use.get("web_search_requests")
+    return getattr(server_tool_use, "web_search_requests", None)
+
+
 def _is_above_128k(tokens: float) -> bool:
     if tokens > 128000:
         return True

litellm/litellm_core_utils/streaming_chunk_builder_utils.py

@@ -588,7 +588,18 @@ def _calculate_usage_per_chunk(
                     hasattr(usage_chunk, "server_tool_use")
                     and usage_chunk.server_tool_use is not None
                 ):
-                    server_tool_use = usage_chunk.server_tool_use
+                    # Coerce dict to ServerToolUse so downstream cost-calc code
+                    # (which accesses .web_search_requests as an attribute)
+                    # doesn't raise AttributeError. Some providers / streaming
+                    # paths leave server_tool_use as a plain dict on the chunk.
+                    if isinstance(usage_chunk.server_tool_use, dict):
+                        server_tool_use = ServerToolUse(**usage_chunk.server_tool_use)
+                    elif isinstance(usage_chunk.server_tool_use, ServerToolUse):
+                        server_tool_use = usage_chunk.server_tool_use
+                    else:
+                        server_tool_use = ServerToolUse.model_validate(
+                            usage_chunk.server_tool_use
+                        )
                 if (
                     usage_chunk_dict["prompt_tokens_details"] is not None
                     and getattr(

litellm/llms/anthropic/chat/transformation.py

@@ -1463,10 +1463,15 @@ def map_openai_params(  # noqa: PLR0915
                 _value = self._map_stop_sequences(value)
                 if _value is not None:
                     optional_params["stop_sequences"] = _value
-            elif param == "temperature":
-                optional_params["temperature"] = value
-            elif param == "top_p":
-                optional_params["top_p"] = value
+            elif param == "temperature" or param == "top_p":
+                AnthropicConfig._apply_sampling_param(
+                    optional_params=optional_params,
+                    model=model,
+                    param=param,
+                    value=value,
+                    drop_params=drop_params,
+                    output_key=param,
+                )
             elif param == "response_format" and isinstance(value, dict):
                 if any(
                     substring in model
@@ -1974,6 +1979,20 @@ def transform_request(
         # Remove internal LiteLLM parameters that should not be sent to Anthropic API
         optional_params.pop("is_vertex_request", None)
 
+        # ``top_k`` is a provider-specific kwarg that bypasses
+        # ``map_openai_params``; gate it here, the single boundary shared by
+        # the direct Anthropic, Bedrock invoke, Vertex, and Azure paths.
+        top_k = optional_params.pop("top_k", None)
+        if top_k is not None:
+            AnthropicConfig._apply_sampling_param(
+                optional_params=optional_params,
+                model=model,
+                param="top_k",
+                value=top_k,
+                drop_params=litellm_params.get("drop_params") is True,
+                output_key="top_k",
+            )
+
         data = {
             "model": model,
             "messages": anthropic_messages,

litellm/llms/anthropic/common_utils.py

@@ -272,19 +272,133 @@ def _is_claude_4_7_model(model: str) -> bool:
         )
 
     @staticmethod
-    def _is_adaptive_thinking_model(model: str) -> bool:
-        """Claude 4.6+ models use adaptive thinking with ``output_config.effort``."""
+    def _supports_sampling_params(model: str) -> bool:
+        """Claude 4.7+ (Opus 4.7/4.8, Fable 5) removed sampling params: the API
+        rejects ``top_p``, ``top_k``, and any ``temperature`` other than 1 with
+        a 400 ("`temperature` is deprecated for this model").
+
+        Driven by the ``supports_sampling_params`` flag in the model map; the
+        name check remains only as a fallback for provider-routed ids whose
+        map entries predate the flag."""
+        flag = AnthropicModelInfo._get_model_capability(
+            model, "supports_sampling_params"
+        )
+        if flag is not None:
+            return flag
+        model_lower = model.lower()
+        return not any(
+            v in model_lower
+            for v in (
+                "fable",
+                "opus-4-7",
+                "opus_4_7",
+                "opus-4.7",
+                "opus_4.7",
+                "opus-4-8",
+                "opus_4_8",
+                "opus-4.8",
+                "opus_4.8",
+            )
+        )
+
+    @staticmethod
+    def _apply_sampling_param(
+        optional_params: dict,
+        model: str,
+        param: str,
+        value: Any,
+        drop_params: bool,
+        output_key: str,
+    ) -> None:
+        """Forward ``temperature``/``top_p``/``top_k`` to
+        ``optional_params[output_key]`` unless the model removed sampling
+        params, in which case drop the param (with drop_params) or raise a
+        clean client-side 400."""
+        if AnthropicModelInfo._supports_sampling_params(model) or (
+            param == "temperature" and value == 1
+        ):
+            optional_params[output_key] = value
+        elif not (litellm.drop_params or drop_params):
+            supported_hint = (
+                "Only temperature=1 is supported. " if param == "temperature" else ""
+            )
+            raise litellm.utils.UnsupportedParamsError(
+                message=(
+                    f"{model} does not support {param}={value}. {supported_hint}"
+                    "To drop unsupported params, set `litellm.drop_params = True`."
+                ),
+                status_code=400,
+            )
+
+    @staticmethod
+    def _model_map_lookup_candidates(model: str) -> List[str]:
+        """Model-map keys to try for ``model``, stripping bedrock/vertex
+        prefixes so a provider-routed Claude still resolves to its entry."""
+        candidates = [model]
+        for prefix in (
+            "bedrock/converse/",
+            "bedrock/invoke/",
+            "bedrock/",
+            "vertex_ai/",
+        ):
+            if model.startswith(prefix):
+                candidates.append(model[len(prefix) :])
+        try:
+            from litellm.llms.bedrock.common_utils import BedrockModelInfo
+
+            base = BedrockModelInfo.get_base_model(model)
+            if base:
+                candidates.append(base)
+                candidates.append(f"bedrock/{base}")
+        except Exception:
+            pass
+        return candidates
+
+    @staticmethod
+    def _get_model_capability(model: str, key: str) -> Optional[bool]:
+        """Read boolean capability ``key`` from the model map, or None when
+        no entry declares it."""
+        try:
+            for cand in AnthropicModelInfo._model_map_lookup_candidates(model):
+                value = litellm.model_cost.get(cand, {}).get(key)
+                if isinstance(value, bool):
+                    return value
+        except Exception:
+            pass
+        return None
+
+    @staticmethod
+    def _supports_model_capability(model: str, key: str) -> bool:
+        """Check a boolean capability ``key`` in the model map.
+
+        Strips bedrock/vertex prefixes so a provider-routed Claude still
+        resolves to the Anthropic model-map entry.
+        """
         from litellm.utils import _supports_factory
 
         try:
             if _supports_factory(
                 model=model,
-                custom_llm_provider=None,
-                key="supports_adaptive_thinking",
+                custom_llm_provider="anthropic",
+                key=key,
             ):
                 return True
         except Exception:
             pass
+        return AnthropicModelInfo._get_model_capability(model, key) is True
+
+    @staticmethod
+    def _is_adaptive_thinking_model(model: str) -> bool:
+        """Claude 4.6+ models use adaptive thinking with ``output_config.effort``.
+
+        Driven by the ``supports_adaptive_thinking`` flag in the model map; the
+        4.6/4.7 name checks remain only as a fallback for provider-routed ids
+        whose map entries predate the flag.
+        """
+        if AnthropicModelInfo._supports_model_capability(
+            model, "supports_adaptive_thinking"
+        ):
+            return True
         return AnthropicModelInfo._is_claude_4_6_model(
             model
         ) or AnthropicModelInfo._is_claude_4_7_model(model)

litellm/llms/anthropic/cost_calculation.py

@@ -7,6 +7,7 @@
 
 from litellm.litellm_core_utils.llm_cost_calc.utils import (
     _get_token_base_cost,
+    _get_web_search_requests,
     _parse_prompt_tokens_details,
     calculate_cache_writing_cost,
     generic_cost_per_token,
@@ -110,11 +111,12 @@ def get_cost_for_anthropic_web_search(
     if model_info is None:
         return 0.0
 
-    if (
-        usage is None
-        or usage.server_tool_use is None
-        or usage.server_tool_use.web_search_requests is None
-    ):
+    if usage is None:
+        return 0.0
+    web_search_requests = _get_web_search_requests(
+        getattr(usage, "server_tool_use", None)
+    )
+    if web_search_requests is None:
         return 0.0
 
     ## Get the cost per web search request
@@ -128,5 +130,5 @@ def get_cost_for_anthropic_web_search(
         return 0.0
 
     ## Calculate the total cost
-    total_cost = cost_per_web_search_request * usage.server_tool_use.web_search_requests
+    total_cost = cost_per_web_search_request * web_search_requests
     return total_cost

litellm/llms/base_llm/responses/transformation.py

@@ -62,6 +62,26 @@ def supports_native_file_search(self) -> bool:
         """
         return False
 
+    def sign_request(
+        self,
+        headers: dict,
+        optional_params: dict,
+        request_data: dict,
+        api_base: str,
+        api_key: Optional[str] = None,
+        model: Optional[str] = None,
+        stream: Optional[bool] = None,
+        fake_stream: Optional[bool] = None,
+    ) -> Tuple[dict, Optional[bytes]]:
+        """Sign the request after the body is finalized.
+
+        Default is a no-op (returns headers unchanged, no signed body). Providers
+        whose endpoint requires request signing (e.g. Bedrock Mantle SigV4)
+        override this and return the signed body bytes so the handler sends those
+        exact bytes.
+        """
+        return headers, None
+
     @abstractmethod
     def get_supported_openai_params(self, model: str) -> list:
         pass