openapi: 3.1.0
info:
  title: NVIDIA NIM Embeddings API
  description: >
    OpenAI-compatible embeddings endpoint backed by NVIDIA NeMo Retriever text
    embedding models. Returns dense float vectors for documents or queries.
    Supports `input_type=passage|query` for asymmetric retrieval and the
    standard `dimensions` parameter on models that allow dimension reduction.
  version: '2026-05-25'
  contact:
    name: NVIDIA Developer Support
    url: https://forums.developer.nvidia.com/c/ai-data-science/nemo-llm-service/
  license:
    name: NVIDIA AI Enterprise License
    url: https://www.nvidia.com/en-us/data-center/products/ai-enterprise/
servers:
  - url: https://integrate.api.nvidia.com
    description: NVIDIA-hosted NIM endpoint
  - url: http://localhost:8000
    description: Self-hosted NIM container default
security:
  - BearerAuth: []
tags:
  - name: Embeddings
    description: Dense vector embedding operations for RAG and semantic search
paths:
  /v1/embeddings:
    post:
      summary: Create An Embedding
      description: Generate embedding vectors for one or more input strings using a NeMo Retriever embedding model.
      operationId: createEmbedding
      tags:
        - Embeddings
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/EmbeddingRequest'
      responses:
        '200':
          description: Embedding response.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/EmbeddingResponse'
        '400':
          description: Invalid request.
        '401':
          description: Missing or invalid API key.
        '429':
          description: Rate limit exceeded.
components:
  securitySchemes:
    BearerAuth:
      type: http
      scheme: bearer
      bearerFormat: nvapi-...
  schemas:
    EmbeddingRequest:
      type: object
      required: [model, input]
      properties:
        model:
          type: string
          description: e.g. `nvidia/llama-3.2-nv-embedqa-1b-v2`, `nvidia/nv-embedqa-e5-v5`, `baai/bge-m3`.
        input:
          oneOf:
            - type: string
            - type: array
              items:
                type: string
        input_type:
          type: string
          enum: [query, passage]
          description: Asymmetric retrieval hint for NV-EmbedQA-style models.
        encoding_format:
          type: string
          enum: [float, base64]
          default: float
        truncate:
          type: string
          enum: [NONE, START, END]
          default: NONE
        dimensions:
          type: integer
          description: Optional output dimensionality for models that support truncation (e.g. Matryoshka models).
        user:
          type: string
    EmbeddingResponse:
      type: object
      properties:
        object:
          type: string
          example: list
        data:
          type: array
          items:
            type: object
            properties:
              object:
                type: string
                example: embedding
              index:
                type: integer
              embedding:
                type: array
                items:
                  type: number
        model:
          type: string
        usage:
          type: object
          properties:
            prompt_tokens:
              type: integer
            total_tokens:
              type: integer