Chroma Query API

openapi: 3.0.3
info:
  title: Chroma Server API (v2) Collections Query API
  description: 'Chroma is an open-source, AI-native vector (embedding) database for LLM, RAG, and semantic-search applications. This OpenAPI describes Chroma''s HTTP/REST v2 API, which is the same interface used by the Python, JavaScript/TypeScript, Rust, and other client libraries. The API is organized around a multi-tenancy hierarchy: tenants contain databases, databases contain collections, and collections contain records (embeddings with documents, metadata, and URIs). The core write and read operations are add, upsert, update, get, query (nearest-neighbor similarity search), and delete.

    ENDPOINTS MODELED: Path structure and the query-collection contract are grounded in the official Chroma reference docs (docs.trychroma.com) and the chroma-core sources. The exact request/response JSON SCHEMAS in this document are MODELED by API Evangelist from the documented client behavior and may differ in field-level detail from Chroma''s own generated openapi.json served at `/openapi.json` on a running server. Treat schemas as representative, not authoritative; verify against your server''s `/openapi.json`.'
  version: '2.0'
  contact:
    name: Chroma
    url: https://www.trychroma.com
  license:
    name: Apache 2.0
    url: https://github.com/chroma-core/chroma/blob/main/LICENSE
servers:
- url: https://api.trychroma.com
  description: Chroma Cloud (managed, serverless)
- url: http://localhost:8000
  description: Local development / self-hosted Chroma server (default port 8000)
security:
- chromaToken: []
tags:
- name: Query
  description: Nearest-neighbor vector similarity search over a collection.
paths:
  /api/v2/tenants/{tenant}/databases/{database}/collections/{collection_id}/query:
    parameters:
    - $ref: '#/components/parameters/Tenant'
    - $ref: '#/components/parameters/Database'
    - $ref: '#/components/parameters/CollectionId'
    post:
      operationId: queryCollection
      tags:
      - Query
      summary: Query a collection (similarity search)
      description: Runs nearest-neighbor vector similarity search over a collection. Supply one or more query embeddings and, optionally, metadata (`where`) and full-text (`where_document`) filters. Returns the closest records with the requested included fields.
      parameters:
      - $ref: '#/components/parameters/Limit'
      - $ref: '#/components/parameters/Offset'
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/QueryRequest'
      responses:
        '200':
          description: Query results.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/QueryResult'
        '401':
          $ref: '#/components/responses/Unauthorized'
        '404':
          $ref: '#/components/responses/NotFound'
components:
  parameters:
    Limit:
      name: limit
      in: query
      required: false
      description: Maximum number of items to return.
      schema:
        type: integer
        minimum: 1
    Offset:
      name: offset
      in: query
      required: false
      description: Number of items to skip for pagination.
      schema:
        type: integer
        minimum: 0
    CollectionId:
      name: collection_id
      in: path
      required: true
      description: The collection UUID.
      schema:
        type: string
    Database:
      name: database
      in: path
      required: true
      description: The database name.
      schema:
        type: string
    Tenant:
      name: tenant
      in: path
      required: true
      description: The tenant name or UUID.
      schema:
        type: string
  responses:
    Unauthorized:
      description: Missing or invalid API token.
      content:
        application/json:
          schema:
            $ref: '#/components/schemas/Error'
    NotFound:
      description: The requested resource was not found.
      content:
        application/json:
          schema:
            $ref: '#/components/schemas/Error'
  schemas:
    QueryRequest:
      type: object
      required:
      - query_embeddings
      properties:
        query_embeddings:
          type: array
          description: One or more query embedding vectors.
          items:
            type: array
            items:
              type: number
              format: float
        n_results:
          type: integer
          description: Number of nearest neighbors to return per query embedding.
          default: 10
        ids:
          type: array
          description: Optional subset of record ids to restrict the search to.
          items:
            type: string
        where:
          type: object
          description: Metadata filter clause.
          additionalProperties: true
        where_document:
          type: object
          description: Full-text / document filter clause.
          additionalProperties: true
        include:
          $ref: '#/components/schemas/Include'
    QueryResult:
      type: object
      description: Results are returned as arrays-of-arrays, one inner array per query embedding.
      properties:
        ids:
          type: array
          items:
            type: array
            items:
              type: string
        distances:
          type: array
          items:
            type: array
            items:
              type: number
              format: float
        documents:
          type: array
          items:
            type: array
            items:
              type: string
              nullable: true
        metadatas:
          type: array
          items:
            type: array
            items:
              type: object
              nullable: true
              additionalProperties: true
        embeddings:
          type: array
          items:
            type: array
            items:
              type: array
              items:
                type: number
                format: float
        uris:
          type: array
          items:
            type: array
            items:
              type: string
              nullable: true
    Include:
      type: array
      description: Which fields to return in results.
      items:
        type: string
        enum:
        - distances
        - documents
        - embeddings
        - metadatas
        - uris
    Error:
      type: object
      properties:
        error:
          type: string
        message:
          type: string
  securitySchemes:
    chromaToken:
      type: apiKey
      in: header
      name: x-chroma-token
      description: Chroma Cloud API key passed in the `x-chroma-token` header. Self-hosted servers can be run open (no auth) or configured with static-token or basic authentication; Chroma Cloud always requires a token.