fix(vector_stores): s3 vectors search router bypass + rag query config drop + ui error swallow

litellm/llms/azure_ai/vector_stores/transformation.py

@@ -19,6 +19,7 @@
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as _LiteLLMLoggingObj
+    from litellm.router import Router
 
     LiteLLMLoggingObj = _LiteLLMLoggingObj
 else:
@@ -115,6 +116,7 @@ def transform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: dict,
         extra_body: dict[str, Any] | None = None,
+        router: "Router | None" = None,
     ) -> tuple[str, dict[str, Any]]:
         """
         Transform search request for Azure AI Search API

litellm/llms/base_llm/vector_store/transformation.py

@@ -17,6 +17,7 @@
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as _LiteLLMLoggingObj
+    from litellm.router import Router
 
     from ..chat.transformation import BaseLLMException as _BaseLLMException
 
@@ -57,6 +58,7 @@ def transform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: dict,
         extra_body: dict[str, Any] | None = None,
+        router: "Router | None" = None,
     ) -> tuple[str, dict]:
         pass
 
@@ -69,6 +71,7 @@ async def atransform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: dict,
         extra_body: dict[str, Any] | None = None,
+        router: "Router | None" = None,
     ) -> tuple[str, dict]:
         """
         Optional async version of transform_search_vector_store_request.
@@ -84,6 +87,7 @@ async def atransform_search_vector_store_request(
             litellm_logging_obj=litellm_logging_obj,
             litellm_params=litellm_params,
             extra_body=extra_body,
+            router=router,
         )
 
     @abstractmethod
@@ -197,6 +201,7 @@ def transform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: Mapping[str, object],
         extra_body: Mapping[str, object] | None = None,
+        router: "Router | None" = None,
     ) -> NoReturn:
         raise NotImplementedError("Direct vector store providers execute the search themselves; no HTTP request shape")
 

litellm/llms/bedrock/vector_stores/transformation.py

@@ -27,6 +27,7 @@
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
+    from litellm.router import Router
 else:
     LiteLLMLoggingObj = Any
 
@@ -196,6 +197,7 @@ def transform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: dict,
         extra_body: dict[str, Any] | None = None,
+        router: "Router | None" = None,
     ) -> tuple[str, dict]:
         if isinstance(query, list):
             query = " ".join(query)

litellm/llms/custom_httpx/llm_http_handler.py

@@ -178,6 +178,7 @@ def _rust_responses_websocket_enabled(
         AnthropicMessagesStreamingResponse,
     )
     from litellm.llms.base_llm.passthrough.transformation import BasePassthroughConfig
+    from litellm.router import Router
     from litellm.types.llms.openai_evals import (
         CancelEvalResponse,
         CancelRunResponse,
@@ -2923,7 +2924,7 @@ async def async_response_api_handler(
         final_response: Final = await self._call_agentic_completion_hooks(
             response=initial_response,
             model=model,
-            messages=(input if isinstance(input, list) else [{"role": "user", "content": input}]),
+            messages=(input if isinstance(input, list) else [{"role": "user", "content": input}]),  # pyright: ignore[reportArgumentType]  # pre-existing mismatch surfaced by the Router import; the hook accepts response input items at runtime
             anthropic_messages_provider_config=responses_api_provider_config,
             anthropic_messages_optional_request_params=response_api_optional_request_params,
             logging_obj=logging_obj,
@@ -5415,7 +5416,7 @@ async def _execute_responses_agentic_plan(
         try:
             response: ResponsesAPIResponse | BaseResponsesAPIStreamingIterator = await litellm.aresponses(
                 model=patch.model or model,
-                input=patch.messages,
+                input=patch.messages,  # pyright: ignore[reportArgumentType]  # pre-existing mismatch surfaced by the Router import; patch messages are valid response input at runtime
                 **optional_params,
                 **kwargs_for_followup,
             )
@@ -9688,6 +9689,7 @@ async def async_vector_store_search_handler(
         timeout: float | httpx.Timeout | None = None,
         client: HTTPHandler | AsyncHTTPHandler | None = None,
         _is_async: bool = False,
+        router: "Router | None" = None,
     ) -> VectorStoreSearchResponse:
         if isinstance(vector_store_provider_config, BaseDirectVectorStoreConfig):
             self._pre_call_direct_vector_store_search(
@@ -9738,6 +9740,7 @@ async def async_vector_store_search_handler(
                 litellm_logging_obj=logging_obj,
                 litellm_params=dict(litellm_params),
                 extra_body=extra_body,
+                router=router,
             )
         else:
             (
@@ -9751,6 +9754,7 @@ async def async_vector_store_search_handler(
                 litellm_logging_obj=logging_obj,
                 litellm_params=dict(litellm_params),
                 extra_body=extra_body,
+                router=router,
             )
         all_optional_params: Final[dict[str, object]] = dict(litellm_params)
         all_optional_params.update(vector_store_search_optional_params or {})
@@ -9802,6 +9806,7 @@ def vector_store_search_handler(
         timeout: float | httpx.Timeout | None = None,
         client: HTTPHandler | AsyncHTTPHandler | None = None,
         _is_async: bool = False,
+        router: "Router | None" = None,
     ) -> VectorStoreSearchResponse | Coroutine[object, object, VectorStoreSearchResponse]:
         if _is_async:
             return self.async_vector_store_search_handler(
@@ -9816,6 +9821,7 @@ def vector_store_search_handler(
                 extra_body=extra_body,
                 timeout=timeout,
                 client=client,
+                router=router,
             )
 
         if isinstance(vector_store_provider_config, BaseDirectVectorStoreConfig):
@@ -9862,6 +9868,7 @@ def vector_store_search_handler(
             litellm_logging_obj=logging_obj,
             litellm_params=dict(litellm_params),
             extra_body=extra_body,
+            router=router,
         )
 
         all_optional_params: Final[dict[str, object]] = dict(litellm_params)

litellm/llms/gemini/vector_stores/transformation.py

@@ -33,6 +33,7 @@
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
+    from litellm.router import Router
 else:
     LiteLLMLoggingObj = Any
 
@@ -168,6 +169,7 @@ def transform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: dict,
         extra_body: Mapping[str, object] | None = None,
+        router: "Router | None" = None,
     ) -> tuple[str, dict]:
         """
         Transform search request to Gemini's generateContent format.

litellm/llms/milvus/vector_stores/transformation.py

@@ -19,6 +19,7 @@
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as _LiteLLMLoggingObj
+    from litellm.router import Router
 
     LiteLLMLoggingObj = _LiteLLMLoggingObj
 else:
@@ -123,6 +124,7 @@ def transform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: dict,
         extra_body: dict[str, Any] | None = None,
+        router: "Router | None" = None,
     ) -> tuple[str, dict[str, Any]]:
         """
         Transform search request for Azure AI Search API

litellm/llms/openai/vector_stores/transformation.py

@@ -21,6 +21,7 @@
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as _LiteLLMLoggingObj
+    from litellm.router import Router
 
     LiteLLMLoggingObj = _LiteLLMLoggingObj
 else:
@@ -99,6 +100,7 @@ def transform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: dict,
         extra_body: dict[str, Any] | None = None,
+        router: "Router | None" = None,
     ) -> tuple[str, dict]:
         encoded_vector_store_id: Final = encode_url_path_segment(vector_store_id, field_name="vector_store_id")
         url: Final = f"{api_base}/{encoded_vector_store_id}/search"

litellm/llms/pg_vector/vector_stores/transformation.py

@@ -8,6 +8,7 @@
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
+    from litellm.router import Router
 else:
     LiteLLMLoggingObj = Any
 
@@ -80,6 +81,7 @@ def transform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: dict,
         extra_body: dict[str, Any] | None = None,
+        router: "Router | None" = None,
     ) -> tuple[str, dict]:
         encoded_vector_store_id: Final = encode_url_path_segment(vector_store_id, field_name="vector_store_id")
         url: Final = f"{api_base}/{encoded_vector_store_id}/search"

litellm/llms/ragflow/vector_stores/transformation.py

@@ -17,6 +17,7 @@
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
+    from litellm.router import Router
 else:
     LiteLLMLoggingObj = Any
 
@@ -92,6 +93,7 @@ def transform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: dict,
         extra_body: dict[str, Any] | None = None,
+        router: "Router | None" = None,
     ) -> tuple[str, dict]:
         """RAGFlow vector stores are management-only, search is not supported."""
         raise NotImplementedError("RAGFlow vector stores support dataset management only, not search/retrieval")

litellm/llms/s3_vectors/vector_stores/transformation.py

@@ -1,8 +1,8 @@
-import re
 from typing import TYPE_CHECKING, Any, Final
 
 import httpx
 
+from litellm.caching._embedding_router import resolve_embedding_router
 from litellm.llms.base_llm.vector_store.transformation import BaseVectorStoreConfig
 from litellm.llms.bedrock.base_aws_llm import BaseAWSLLM
 from litellm.types.router import GenericLiteLLMParams
@@ -18,6 +18,7 @@
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
+    from litellm.router import Router
 else:
     LiteLLMLoggingObj = Any
 
@@ -58,13 +59,20 @@ def validate_environment(self, headers: dict, litellm_params: GenericLiteLLMPara
         return headers
 
     def get_complete_url(self, api_base: str | None, litellm_params: dict) -> str:
-        aws_region_name: Final = litellm_params.get("aws_region_name")
-        if not aws_region_name:
-            raise ValueError("aws_region_name is required for S3 Vectors")
-        if not re.match(r"^[a-z][a-z0-9-]*$", aws_region_name):
-            raise ValueError("Invalid aws_region_name format")
+        # Resolve region the same way the ingestion path does:
+        # dynamic param -> AWS_REGION_NAME -> AWS_REGION -> default (us-west-2)
+        aws_region_name: Final = self.get_aws_region_name_for_non_llm_api_calls(litellm_params.get("aws_region_name"))
         return f"https://s3vectors.{aws_region_name}.api.aws"
 
+    def _resolve_query_embedding_router(self, embedding_model: str, router: "Router | None") -> "Router | None":
+        """Return the router iff it serves ``embedding_model`` as a deployment."""
+        if router is None:
+            return None
+        model_list: Final = [
+            dict(m) for m in (router.get_model_list() or ())
+        ]  # mutable-ok: resolve_embedding_router requires list[dict]
+        return resolve_embedding_router(embedding_model=embedding_model, llm_router=router, llm_model_list=model_list)
+
     def transform_search_vector_store_request(
         self,
         vector_store_id: str,
@@ -74,6 +82,7 @@ def transform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: dict,
         extra_body: dict[str, Any] | None = None,
+        router: "Router | None" = None,
     ) -> tuple[str, dict]:
         """Sync version - generates embedding synchronously."""
         # For S3 Vectors, vector_store_id should be in format: bucket_name:index_name
@@ -99,10 +108,16 @@ def transform_search_vector_store_request(
 
         # Generate embedding for the query
         embedding_model: Final = litellm_params.get("embedding_model", "text-embedding-3-small")
+        embedding_router: Final = self._resolve_query_embedding_router(embedding_model=embedding_model, router=router)
 
         import litellm as litellm_module
 
-        embedding_response: Final = litellm_module.embedding(model=embedding_model, input=[query])
+        embedding_input: Final = [query]  # mutable-ok: the embedding API takes list input
+        embedding_response: Final = (
+            embedding_router.embedding(model=embedding_model, input=embedding_input)
+            if embedding_router is not None
+            else litellm_module.embedding(model=embedding_model, input=embedding_input)
+        )
         query_embedding: Final = embedding_response.data[0]["embedding"]
 
         url: Final = f"{api_base}/QueryVectors"
@@ -128,6 +143,7 @@ async def atransform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: dict,
         extra_body: dict[str, Any] | None = None,
+        router: "Router | None" = None,
     ) -> tuple[str, dict]:
         """Async version - generates embedding asynchronously."""
         # For S3 Vectors, vector_store_id should be in format: bucket_name:index_name
@@ -153,10 +169,16 @@ async def atransform_search_vector_store_request(
 
         # Generate embedding for the query asynchronously
         embedding_model: Final = litellm_params.get("embedding_model", "text-embedding-3-small")
+        embedding_router: Final = self._resolve_query_embedding_router(embedding_model=embedding_model, router=router)
 
         import litellm as litellm_module
 
-        embedding_response: Final = await litellm_module.aembedding(model=embedding_model, input=[query])
+        embedding_input: Final = [query]  # mutable-ok: the embedding API takes list input
+        embedding_response: Final = (
+            await embedding_router.aembedding(model=embedding_model, input=embedding_input)
+            if embedding_router is not None
+            else await litellm_module.aembedding(model=embedding_model, input=embedding_input)
+        )
         query_embedding: Final = embedding_response.data[0]["embedding"]
 
         url: Final = f"{api_base}/QueryVectors"

litellm/llms/vertex_ai/vector_stores/rag_api/transformation.py

@@ -21,6 +21,7 @@
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as _LiteLLMLoggingObj
+    from litellm.router import Router
 
     LiteLLMLoggingObj = _LiteLLMLoggingObj
 else:
@@ -161,6 +162,7 @@ def transform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: dict,
         extra_body: Mapping[str, object] | None = None,
+        router: "Router | None" = None,
     ) -> tuple[str, dict[str, object]]:
         """
         Transform search request for Vertex AI RAG API

litellm/llms/vertex_ai/vector_stores/search_api/transformation.py

@@ -25,6 +25,7 @@
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as _LiteLLMLoggingObj
+    from litellm.router import Router
 
     LiteLLMLoggingObj = _LiteLLMLoggingObj
 else:
@@ -245,6 +246,7 @@ def transform_search_vector_store_request(
         litellm_logging_obj: LiteLLMLoggingObj,
         litellm_params: dict,
         extra_body: Mapping[str, object] | None = None,
+        router: "Router | None" = None,
     ) -> tuple[str, dict[str, object]]:
         """
         Transform a search request for the Vertex AI Search (Discovery Engine) API.