Question Answering
POST/v1/vector_stores/question-answering
Authorization
Authorizationstringrequired
Bearer token for API authentication. Format: `Bearer YOUR_API_KEY`
Request Body
querystring
Question to answer. If not provided, the question will be extracted from the passed messages.
Constraints
•Minimum length: 1
vector_store_identifiersarray | null
IDs or names of vector stores to search
top_kintegerdefault:
10Number of results to return
Constraints
•Minimum: 1
file_idsarray | array
Optional list of file IDs to filter chunks by (inclusion filter)
Constraints
•Minimum items: 2•Maximum items: 2
streambooleandefault:
falseWhether to stream the answer
Response Body
answerstringrequired
The answer generated by the LLM
Search Files
Perform semantic search across complete vector store files. This endpoint searches through vector store files using semantic similarity matching. Unlike chunk search, it returns complete matching files rather than individual chunks. Supports complex search queries with filters and returns relevance-scored results.
Create Embeddings
Create embeddings for text or images using the specified model, encoding format, and normalization.
Last updated: August 19, 2025