Create a vector store
Creates a new vector store for storing and searching through document embeddings.
Authorizations
Gateway auth: send Authorization: Bearer <mka1-api-key>. For multi-user server-side integrations, you can also send X-On-Behalf-Of: <external-user-id>.
Headers
Optional external end-user identifier forwarded by the API gateway.
Body
Request body for creating a vector store.
The name of the vector store.
A description for the vector store. Can be used to describe the vector store's purpose.
A list of File IDs that the vector store should use. Useful for tools like file_search that can access files. At most 500 per request.
500The expiration policy for a vector store.
The chunking strategy used to chunk the file(s). If not set, will use the auto strategy. Only applicable if file_ids is non-empty.
- Option 1
- Option 2
Set of 16 key-value pairs that can be attached to an object. Keys are strings with a maximum length of 64 characters. Values are strings with a maximum length of 512 characters.
The embedding model to use. Defaults to the auto-configured model if not specified.
The number of dimensions for the embedding vectors. Only supported for models with flexible dimensions. If not specified, uses the model's default dimensions.
-9007199254740991 < x <= 9007199254740991Retrieval mode, frozen at creation (cannot be changed later). 'vector' (default): standard vector similarity search. 'graph': GraphRAG — entities and relations are extracted from every chunk at ingest (metered against your usage) and search traverses the knowledge graph.
vector, graph Model used for entity/relation extraction on graph stores (ingest-time triplet extraction and query-entity extraction). Defaults to the auto-configured model, resolved at creation time — same contract as embedding_model. Only valid when retrieval_mode is 'graph'.
Graph expansion depth for graph-mode queries (1-4, engine default 2). Only valid for graph stores.
1 <= x <= 4Response
OK
A vector store is a collection of processed files that can be used by the file_search tool.
The identifier, which can be referenced in API endpoints.
The object type, which is always 'vector_store'.
The Unix timestamp (in seconds) for when the vector store was created.
-9007199254740991 <= x <= 9007199254740991The name of the vector store.
The total number of bytes used by the files in the vector store.
-9007199254740991 <= x <= 9007199254740991The embedding model used for this vector store. Resolved at creation time from the requested or auto model. Null for legacy vector stores.
The number of dimensions for the embedding vectors in this vector store. Null for legacy vector stores.
0 < x <= 9007199254740991Retrieval mode, frozen at creation. 'vector': standard vector similarity search. 'graph': GraphRAG — entities and relations are extracted at ingest and queries traverse the knowledge graph. 'hybrid' is reserved.
vector, hybrid, graph File processing status counts.
The status of the vector store. 'expired' means the store has expired, 'in_progress' means files are still being processed, 'completed' indicates that the vector store is ready for use.
expired, in_progress, completed The Unix timestamp (in seconds) for when the vector store was last active.
-9007199254740991 <= x <= 9007199254740991Set of 16 key-value pairs that can be attached to an object. Keys are strings with a maximum length of 64 characters. Values are strings with a maximum length of 512 characters, booleans, or numbers.
The Unix timestamp (in seconds) for when the vector store was last used.
-9007199254740991 <= x <= 9007199254740991The model used for entity/relation extraction on graph stores. Null for non-graph stores.
Graph expansion depth for graph-mode queries. Null for non-graph stores.
1 <= x <= 4The expiration policy for a vector store.
The Unix timestamp (in seconds) for when the vector store will expire.
-9007199254740991 <= x <= 9007199254740991A description for the vector store. Can be used to describe the vector store's purpose.