Vectors

For advanced use cases, you can directly manage vectors without going through the document ingestion pipeline. This is useful for custom embeddings or integrating with external embedding services.

Upsert Vectors

POST /v1/collections/{collection_name}/vectors/upsert

Insert or update vectors directly.

Request body:

{
  "vectors": [
    {
      "id": "custom_001",
      "embedding": [0.1, 0.2, 0.3],
      "text": "Custom text content",
      "metadata": { "source": "external" }
    }
  ]
}

Field	Type	Required	Description
`vectors`	array	yes	List of vector entries
`vectors[].id`	string	yes	Unique vector ID
`vectors[].embedding`	array[float]	yes	Must match collection dimension
`vectors[].text`	string	no	Associated text content
`vectors[].metadata`	object	no	Arbitrary metadata

Response 200:

{ "status": "ok", "upserted": 1 }

Errors:

400 — Dimension mismatch, invalid vectors
404 — Collection not found

Delete Vectors

POST /v1/collections/{collection_name}/vectors/delete

Delete vectors by their IDs.

Request body:

{ "ids": ["custom_001", "custom_002"] }

Response 200:

{ "status": "ok", "deleted": 2 }

Errors: 404 — Collection not found

Example: Custom Embeddings

# Upsert vectors with your own embeddings
curl -X POST http://localhost:6100/v1/collections/custom/vectors/upsert \
  -H "Authorization: Bearer $BIGRAG_API_SECRET" \
  -H "Content-Type: application/json" \
  -d '{
    "vectors": [
      {
        "id": "vec_001",
        "embedding": [0.1, 0.2, 0.3],
        "text": "Custom content",
        "metadata": {"source": "external"}
      }
    ]
  }'

# Delete vectors
curl -X POST http://localhost:6100/v1/collections/custom/vectors/delete \
  -H "Authorization: Bearer $BIGRAG_API_SECRET" \
  -H "Content-Type: application/json" \
  -d '{"ids": ["vec_001"]}'

The embedding dimension must match the collection's configured dimension. A dimension mismatch will return a 400 error.

Upsert Vectors

Delete Vectors

Example: Custom Embeddings

On this page