Implementation:BerriAI Litellm Vector Store Files API

Property	Value
sources	litellm/vector_store_files/main.py
domains	Vector Stores, File Management, CRUD, OpenAI
last_updated	2026-02-15 16:00 GMT

Overview

The Vector Store Files API module provides full CRUD operations for managing files within vector stores, including creating, listing, retrieving, retrieving content, updating attributes, and deleting vector store files with automatic registry credential integration.

Description

This module implements six sync/async function pairs for vector store file management: create/acreate, list/alist, retrieve/aretrieve, retrieve_content/aretrieve_content, update/aupdate, and delete/adelete. All functions use the @client decorator and default to "openai" as the provider via the _ensure_provider() helper. Each function integrates with the vector store registry through _prepare_registry_credentials(), which automatically pulls stored credentials for the given vector store ID when litellm.vector_store_registry is available. The module uses VectorStoreFileRequestUtils for parameter extraction, ProviderConfigManager.get_provider_vector_store_files_config() for provider config loading, and delegates HTTP calls to specialized handlers on BaseLLMHTTPHandler. File attributes support typed values (str, int, float, bool) and can be set during creation or updated via the update endpoint.

Usage

Import this module when you need to manage individual files within a vector store, including uploading files, listing files with pagination, retrieving file metadata or content, updating file attributes for filtering, and deleting files.

Code Reference

Source Location

Property	Value
Repository	github.com/BerriAI/litellm
File	litellm/vector_store_files/main.py
Lines	781
Module	litellm.vector_store_files.main

Signature

@client
def create(
    *, vector_store_id: str, file_id: str,
    attributes: Optional[VectorStoreFileAttributes] = None,
    chunking_strategy: Optional[Dict[str, Any]] = None,
    extra_headers=None, extra_query=None, extra_body=None,
    timeout=None, custom_llm_provider=None, **kwargs,
) -> Union[VectorStoreFileObject, Coroutine[Any, Any, VectorStoreFileObject]]

@client
def list(*, vector_store_id: str, after=None, before=None, filter=None,
    limit=None, order=None, ...) -> VectorStoreFileListResponse

@client
def retrieve(*, vector_store_id: str, file_id: str, ...) -> VectorStoreFileObject

@client
def retrieve_content(*, vector_store_id: str, file_id: str, ...) -> VectorStoreFileContentResponse

@client
def update(*, vector_store_id: str, file_id: str,
    attributes: VectorStoreFileAttributes, ...) -> VectorStoreFileObject

@client
def delete(*, vector_store_id: str, file_id: str, ...) -> VectorStoreFileDeleteResponse

Import

from litellm.vector_store_files.main import (
    create, acreate,
    list, alist,
    retrieve, aretrieve,
    retrieve_content, aretrieve_content,
    update, aupdate,
    delete, adelete,
)

I/O Contract

Inputs

Parameter	Type	Required	Description
vector_store_id	str	Yes	The ID of the vector store containing the files
file_id	str	For create/retrieve/update/delete	The ID of the file to operate on
attributes	VectorStoreFileAttributes	For update	Key-value pairs (str, int, float, bool values)
chunking_strategy	Optional[Dict[str, Any]]	No	Chunking strategy for file processing
after	Optional[str]	No	Pagination cursor for list
before	Optional[str]	No	Reverse pagination cursor for list
filter	Optional[str]	No	Filter for list results
limit	Optional[int]	No	Number of items per page
custom_llm_provider	Optional[str]	No	Provider; defaults to "openai"

Outputs

Function	Return Type	Description
create	VectorStoreFileObject	The created file object with metadata
list	VectorStoreFileListResponse	Paginated list of files
retrieve	VectorStoreFileObject	The retrieved file object
retrieve_content	VectorStoreFileContentResponse	The file's parsed content
update	VectorStoreFileObject	The updated file object
delete	VectorStoreFileDeleteResponse	Deletion confirmation

Usage Examples

import litellm

# Create a file in a vector store
file_obj = litellm.vector_store_files.create(
    vector_store_id="vs_abc123",
    file_id="file_xyz789",
    attributes={"category": "documentation", "priority": 1},
)
print(f"File: {file_obj.id}, Status: {file_obj.status}")

import litellm

# List files in a vector store
files = litellm.vector_store_files.list(
    vector_store_id="vs_abc123",
    limit=10,
)
for f in files.data:
    print(f"File: {f.id}")

# Update file attributes
updated = litellm.vector_store_files.update(
    vector_store_id="vs_abc123",
    file_id="file_xyz789",
    attributes={"category": "updated_docs", "reviewed": True},
)

Related Pages

• BerriAI_Litellm_Vector_Stores_API -- Vector store create and search operations that contain these files
• BerriAI_Litellm_Vector_Store_Registry -- Registry that provides credentials for vector store file operations
• BerriAI_Litellm_Containers_API -- Container file management with similar CRUD patterns