chore: litellm oss staging

README.md

@@ -345,6 +345,7 @@ curl -X POST 'http://0.0.0.0:4000/v1/chat/completions' \
 | [OVHCloud AI Endpoints (`ovhcloud`)](https://docs.litellm.ai/docs/providers/ovhcloud) | ✅ | ✅ | ✅ |  |  |  |  |  |  |  |
 | [Perplexity AI (`perplexity`)](https://docs.litellm.ai/docs/providers/perplexity) | ✅ | ✅ | ✅ |  |  |  |  |  |  |  |
 | [Petals (`petals`)](https://docs.litellm.ai/docs/providers/petals) | ✅ | ✅ | ✅ |  |  |  |  |  |  |  |
+| [Pinstripes (`pinstripes`)](https://docs.litellm.ai/docs/providers/pinstripes) | ✅ | ✅ | ✅ |  |  |  |  |  |  |  |
 | [Predibase (`predibase`)](https://docs.litellm.ai/docs/providers/predibase) | ✅ | ✅ | ✅ |  |  |  |  |  |  |  |
 | [Recraft (`recraft`)](https://docs.litellm.ai/docs/providers/recraft) |  |  |  |  | ✅ |  |  |  |  |  |
 | [Replicate (`replicate`)](https://docs.litellm.ai/docs/providers/replicate) | ✅ | ✅ | ✅ |  |  |  |  |  |  |  |

enterprise/litellm_enterprise/proxy/hooks/managed_files.py

@@ -1,9 +1,9 @@
 # What is this?
 ## This hook is used to check for LiteLLM managed files in the request body, and replace them with model-specific file id
 
-import asyncio
 import base64
 import json
+from types import MappingProxyType
 from typing import TYPE_CHECKING, Any, Dict, List, Literal, Optional, Union, cast
 
 from fastapi import HTTPException
@@ -1472,8 +1472,8 @@ async def _check_file_deletion_allowed(self, file_id: str) -> None:
                 error_message += f" (showing {MAX_BATCHES_IN_ERROR} most recent): {', '.join(batch_statuses)}. "
 
             error_message += (
-                f"To delete this file before complete cost tracking, please delete or cancel the referencing batch(es) first. "
-                f"Alternatively, wait for all batches to complete and for cost to be computed (batch_processed=true)."
+                "To delete this file before complete cost tracking, please delete or cancel the referencing batch(es) first. "
+                "Alternatively, wait for all batches to complete and for cost to be computed (batch_processed=true)."
             )
 
             # Record blocked deletion metric
@@ -1550,9 +1550,22 @@ async def afile_content(
 
         if specific_model_file_id_mapping:
             exception_dict = {}
-            for model_id, file_id in specific_model_file_id_mapping.items():
+            for model_id, provider_file_id in specific_model_file_id_mapping.items():
                 try:
-                    return await llm_router.afile_content(model=model_id, file_id=file_id, **data)  # type: ignore
+                    # Cloud-storage providers (e.g. Bedrock S3) validate file ids
+                    # against the deployment's configured bucket, which they only
+                    # trust from this immutable server-side snapshot, never from
+                    # request params.
+                    credentials = llm_router.get_deployment_credentials_with_provider(
+                        model_id=model_id
+                    )
+                    if credentials is not None:
+                        data["_litellm_internal_model_credentials"] = cast(
+                            Dict, MappingProxyType(dict(credentials))
+                        )
+                    else:
+                        data.pop("_litellm_internal_model_credentials", None)
+                    return await llm_router.afile_content(model=model_id, file_id=provider_file_id, **data)  # type: ignore
                 except Exception as e:
                     exception_dict[model_id] = str(e)
             raise Exception(

litellm/caching/gcs_cache.py

@@ -5,6 +5,7 @@
 import json
 import asyncio
 from typing import Optional
+from urllib.parse import quote
 
 from litellm._logging import print_verbose, verbose_logger
 from litellm.integrations.gcs_bucket.gcs_bucket_base import GCSBucketBase
@@ -48,7 +49,7 @@ def set_cache(self, key, value, **kwargs):
             headers = self._construct_headers()
             object_name = self.key_prefix + key
             bucket_name = self.bucket_name
-            url = f"https://storage.googleapis.com/upload/storage/v1/b/{bucket_name}/o?uploadType=media&name={object_name}"
+            url = f"https://storage.googleapis.com/upload/storage/v1/b/{bucket_name}/o?uploadType=media&name={quote(object_name, safe='')}"
             data = json.dumps(value)
             self.sync_client.post(url=url, data=data, headers=headers)
         except Exception as e:
@@ -59,7 +60,7 @@ async def async_set_cache(self, key, value, **kwargs):
             headers = self._construct_headers()
             object_name = self.key_prefix + key
             bucket_name = self.bucket_name
-            url = f"https://storage.googleapis.com/upload/storage/v1/b/{bucket_name}/o?uploadType=media&name={object_name}"
+            url = f"https://storage.googleapis.com/upload/storage/v1/b/{bucket_name}/o?uploadType=media&name={quote(object_name, safe='')}"
             data = json.dumps(value)
             await self.async_client.post(url=url, data=data, headers=headers)
         except Exception as e:
@@ -72,7 +73,7 @@ def get_cache(self, key, **kwargs):
             headers = self._construct_headers()
             object_name = self.key_prefix + key
             bucket_name = self.bucket_name
-            url = f"https://storage.googleapis.com/storage/v1/b/{bucket_name}/o/{object_name}?alt=media"
+            url = f"https://storage.googleapis.com/storage/v1/b/{bucket_name}/o/{quote(object_name, safe='')}?alt=media"
             response = self.sync_client.get(url=url, headers=headers)
             if response.status_code == 200:
                 cached_response = json.loads(response.text)
@@ -91,7 +92,7 @@ async def async_get_cache(self, key, **kwargs):
             headers = self._construct_headers()
             object_name = self.key_prefix + key
             bucket_name = self.bucket_name
-            url = f"https://storage.googleapis.com/storage/v1/b/{bucket_name}/o/{object_name}?alt=media"
+            url = f"https://storage.googleapis.com/storage/v1/b/{bucket_name}/o/{quote(object_name, safe='')}?alt=media"
             response = await self.async_client.get(url=url, headers=headers)
             if response.status_code == 200:
                 return json.loads(response.text)

litellm/constants.py

@@ -802,6 +802,7 @@
     "https://api.inference.wandb.ai/v1",
     "https://api.clarifai.com/v2/ext/openai/v1",
     "https://api.libertai.io/v1",
+    "https://pinstripes.io/v1",
 ]
 
 
@@ -865,6 +866,7 @@
     "clarifai",
     "docker_model_runner",
     "ragflow",
+    "pinstripes",  # Pinstripes - JSON-configured provider
 ]
 openai_text_completion_compatible_providers: List = (
     [  # providers that support `/v1/completions`

litellm/litellm_core_utils/cloud_storage_security.py

@@ -15,8 +15,23 @@
     BEDROCK_MANAGED_S3_UPLOAD_PREFIX,
     BEDROCK_MANAGED_S3_OUTPUT_PREFIX,
 )
+MANAGED_CLOUD_STORAGE_SCHEMES = ("s3://", "gs://")
 _MAPPING_PROXY_TYPE: type = type(MappingProxyType({}))
 
+
+def is_managed_cloud_storage_uri(file_id: str) -> bool:
+    """
+    True if file_id is a raw cloud-storage object URI (e.g. ``s3://bucket/key``).
+
+    These are internal provider artifacts. On the multi-tenant proxy they must be
+    retrieved through their managed unified file id so owner/team access is enforced;
+    a raw URI supplied by a caller bypasses that check.
+    """
+    return isinstance(file_id, str) and file_id.startswith(
+        MANAGED_CLOUD_STORAGE_SCHEMES
+    )
+
+
 _SAFE_OBJECT_COMPONENT_PATTERN = re.compile(r"[^A-Za-z0-9._-]+")
 
 

litellm/litellm_core_utils/get_llm_provider_logic.py

@@ -388,6 +388,9 @@ def get_llm_provider(
                     elif endpoint == "https://api.inference.wandb.ai/v1":
                         custom_llm_provider = "wandb"
                         dynamic_api_key = get_secret_str("WANDB_API_KEY")
+                    elif endpoint == "https://pinstripes.io/v1":
+                        custom_llm_provider = "pinstripes"
+                        dynamic_api_key = get_secret_str("PINSTRIPES_API_KEY")
 
                     if api_base is not None and not isinstance(api_base, str):
                         raise Exception(
@@ -641,7 +644,7 @@ def _get_openai_compatible_provider_info(
             api_base,
             dynamic_api_key,
         ) = litellm.BedrockMantleChatConfig()._get_openai_compatible_provider_info(
-            api_base, api_key, litellm_params=litellm_params
+            api_base, api_key, litellm_params=litellm_params, model=model
         )
     elif custom_llm_provider == "nvidia_nim":
         # nvidia_nim is openai compatible, we just need to set this to custom_openai and have the api_base be https://api.endpoints.anyscale.com/v1

litellm/litellm_core_utils/health_check_helpers.py

@@ -44,8 +44,8 @@ async def ahealth_check_wildcard_models(
         model_params["litellm_logging_obj"] = litellm_logging_obj
         model_params["fallbacks"] = fallback_models
         model_params["max_tokens"] = model_params.get(
-            "max_tokens", 10
-        )  # gpt-5-nano throws errors for max_tokens=1
+            "max_tokens", 16
+        )  # GPT-5 models require max_output_tokens >= 16
         await acompletion(**model_params)
         return {}
 

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

@@ -859,7 +859,17 @@ def translate_anthropic_tools_to_openai(
         """
         new_tools: List[ChatCompletionToolParam] = []
         tool_name_mapping: Dict[str, str] = {}
-        mapped_tool_params = ["name", "input_schema", "description", "cache_control"]
+        # "type" is the Anthropic tool type (e.g. "custom"); it must not be
+        # merged into the OpenAI function `parameters` schema below, or it
+        # overwrites the real parameters.type ("object") and the provider
+        # rejects the request. See #30557.
+        mapped_tool_params = [
+            "name",
+            "input_schema",
+            "description",
+            "cache_control",
+            "type",
+        ]
 
         for idx, tool in enumerate(tools):
             # Check if this is an Anthropic-native tool that should be kept as-is

litellm/llms/bedrock/files/handler.py

@@ -1,8 +1,6 @@
 import asyncio
-import base64
-import os
-from types import MappingProxyType
-from typing import Any, Coroutine, Mapping, Optional, Tuple, Union, cast
+from collections.abc import Mapping
+from typing import Any, Coroutine, Optional, Tuple, Union
 
 import httpx
 
@@ -17,7 +15,6 @@
     FileContentRequest,
     HttpxBinaryResponseContent,
 )
-from litellm.types.utils import SpecialEnums
 
 from ..base_aws_llm import BaseAWSLLM
 
@@ -37,40 +34,9 @@ def __init__(self):
         )
 
     def _extract_s3_uri_from_file_id(self, file_id: str) -> str:
-        """
-        Extract S3 URI from encoded file ID.
-
-        The file ID can be in two formats:
-        1. Base64-encoded unified file ID containing: llm_output_file_id,s3://bucket/path
-        2. Direct S3 URI: s3://bucket/litellm-managed-prefix/path
-
-        Args:
-            file_id: Encoded file ID or direct S3 URI
+        from .transformation import extract_s3_uri_from_file_id
 
-        Returns:
-            S3 URI (e.g., "s3://bucket-name/path/to/file")
-        """
-        # First, try to decode if it's a base64-encoded unified file ID
-        try:
-            # Add padding if needed
-            padded = file_id + "=" * (-len(file_id) % 4)
-            decoded = base64.urlsafe_b64decode(padded).decode()
-
-            # Check if it's a unified file ID format
-            if decoded.startswith(SpecialEnums.LITELM_MANAGED_FILE_ID_PREFIX.value):
-                # Extract llm_output_file_id from the decoded string
-                if "llm_output_file_id," in decoded:
-                    s3_uri = decoded.split("llm_output_file_id,")[1].split(";")[0]
-                    return s3_uri
-        except Exception:
-            pass
-
-        # If not base64 encoded or doesn't contain llm_output_file_id, accept only
-        # explicit S3 URIs. Bucket and key validation happens before any S3 call.
-        if file_id.startswith("s3://"):
-            return file_id
-
-        raise ValueError("file_id must be a managed LiteLLM S3 file id")
+        return extract_s3_uri_from_file_id(file_id)
 
     def _parse_s3_uri(
         self,
@@ -95,26 +61,12 @@ def _parse_s3_uri(
             allow_legacy_cloud_file_ids=allow_legacy_cloud_file_ids,
         )
 
-    def _get_configured_s3_bucket_name(self, litellm_params: dict) -> str:
-        trusted_model_credentials = litellm_params.get(
-            "_litellm_internal_model_credentials"
-        )
-        bucket_name = None
-        if isinstance(trusted_model_credentials, type(MappingProxyType({}))):
-            trusted_model_credentials_mapping = cast(
-                Mapping[str, Any], trusted_model_credentials
-            )
-            candidate_bucket_name = trusted_model_credentials_mapping.get(
-                "s3_bucket_name"
-            )
-            if isinstance(candidate_bucket_name, str):
-                bucket_name = candidate_bucket_name
-        bucket_name = bucket_name or os.getenv("AWS_S3_BUCKET_NAME")
-        if not bucket_name:
-            raise ValueError(
-                "S3 bucket_name is required. Set 's3_bucket_name' in proxy config or AWS_S3_BUCKET_NAME for Bedrock file content retrieval."
-            )
-        return bucket_name
+    def _get_configured_s3_bucket_name(
+        self, litellm_params: Mapping[str, object]
+    ) -> str:
+        from .transformation import get_configured_s3_bucket_name
+
+        return get_configured_s3_bucket_name(litellm_params)
 
     async def afile_content(
         self,