Triton Inference Server Model Repository API

openapi: 3.1.0
info:
  title: Triton Inference Server NVIDIA Triton Inference Server HTTP/REST CUDA Shared Memory Model Repository API
  description: RESTful API for the NVIDIA Triton Inference Server, implementing the KServe V2 inference protocol with Triton-specific extensions. Provides endpoints for model inference, health checks, server and model metadata, model repository management, statistics, tracing, logging, and system shared memory management.
  version: '2.0'
  contact:
    name: NVIDIA Triton Team
    url: https://github.com/triton-inference-server/server
    email: triton@nvidia.com
  license:
    name: BSD 3-Clause
    url: https://github.com/triton-inference-server/server/blob/main/LICENSE
servers:
- url: http://localhost:8000
  description: Triton HTTP endpoint (default)
- url: http://{host}:{port}
  description: Custom Triton HTTP endpoint
  variables:
    host:
      default: localhost
      description: Triton server hostname or IP
    port:
      default: '8000'
      description: Triton HTTP port
tags:
- name: Model Repository
  description: Model repository management operations
paths:
  /v2/repository/index:
    post:
      operationId: repositoryIndex
      summary: Triton Inference Server List Models in the Repository
      description: Retrieve the index of all models available in the model repository. Can optionally filter to show only models that are ready. This is a Triton extension to the KServe protocol.
      tags:
      - Model Repository
      requestBody:
        required: false
        content:
          application/json:
            schema:
              type: object
              properties:
                ready:
                  type: boolean
                  description: If true, only return models that are ready for inference
      responses:
        '200':
          description: Repository index returned successfully
          content:
            application/json:
              schema:
                type: array
                items:
                  $ref: '#/components/schemas/RepositoryIndexEntry'
        '400':
          description: Error retrieving repository index
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ErrorResponse'
  /v2/repository/models/{model_name}/load:
    post:
      operationId: modelLoad
      summary: Triton Inference Server Load or Reload a Model
      description: Request that a model be loaded into Triton, or reloaded if it is already loaded. Optionally provide model configuration overrides as parameters. This is a Triton extension to the KServe protocol.
      tags:
      - Model Repository
      parameters:
      - $ref: '#/components/parameters/modelName'
      requestBody:
        required: false
        content:
          application/json:
            schema:
              type: object
              properties:
                parameters:
                  type: object
                  description: Optional parameters for model loading, including config overrides and file content overrides.
                  additionalProperties:
                    oneOf:
                    - type: string
                    - type: boolean
                    - type: integer
      responses:
        '200':
          description: Model loaded successfully
        '400':
          description: Error loading model
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ErrorResponse'
  /v2/repository/models/{model_name}/unload:
    post:
      operationId: modelUnload
      summary: Triton Inference Server Unload a Model
      description: Request that a model be unloaded from Triton. Once unloaded the model will no longer be available for inference. This is a Triton extension to the KServe protocol.
      tags:
      - Model Repository
      parameters:
      - $ref: '#/components/parameters/modelName'
      requestBody:
        required: false
        content:
          application/json:
            schema:
              type: object
              properties:
                parameters:
                  type: object
                  description: Optional parameters for model unloading
                  additionalProperties:
                    oneOf:
                    - type: string
                    - type: boolean
                    - type: integer
      responses:
        '200':
          description: Model unloaded successfully
        '400':
          description: Error unloading model
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ErrorResponse'
components:
  parameters:
    modelName:
      name: model_name
      in: path
      required: true
      description: Name of the model
      schema:
        type: string
  schemas:
    RepositoryIndexEntry:
      type: object
      properties:
        name:
          type: string
          description: Model name
        version:
          type: string
          description: Model version
        state:
          type: string
          description: Current state of the model
          enum:
          - READY
          - UNAVAILABLE
          - LOADING
          - UNLOADING
        reason:
          type: string
          description: Reason for the current state if not READY
    ErrorResponse:
      type: object
      properties:
        error:
          type: string
          description: Error message describing what went wrong
externalDocs:
  description: Triton Inference Server Protocol Documentation
  url: https://github.com/triton-inference-server/server/blob/main/docs/protocol/README.md