Google TensorFlow
TensorFlow Serving REST API

TensorFlow Serving provides a REST API for serving trained TensorFlow models in production environments. The API supports model prediction (inference), classification, and regression requests against deployed models. It allows specifying model names and versions, and returns predictions in JSON format. TensorFlow Serving handles model lifecycle management, versioning, and concurrent request processing.
Documentation GitHub OpenAPI
OpenAPI Specification

openapi: 3.1.0
info:
  title: Google TensorFlow TensorFlow Serving REST API
  description: >-
    TensorFlow Serving exposes a REST API for running inference on trained
    TensorFlow models. It supports predict, classify, and regress endpoints
    with model versioning and lifecycle management.
  version: '1'
  contact:
    name: TensorFlow Community
    url: https://www.tensorflow.org/community
  license:
    name: Apache 2.0
    url: https://www.apache.org/licenses/LICENSE-2.0
externalDocs:
  description: TensorFlow Serving REST API Documentation
  url: https://www.tensorflow.org/tfx/serving/api_rest
servers:
  - url: http://localhost:8501
    description: Default TensorFlow Serving REST endpoint
tags:
  - name: Model Status
    description: Model metadata and status operations
  - name: Prediction
    description: Model inference operations
paths:
  /v1/models/{model_name}:
    get:
      operationId: getModelStatus
      summary: Google TensorFlow Get model status
      description: Returns the status of a model including available versions.
      tags:
        - Model Status
      parameters:
        - $ref: '#/components/parameters/model_name'
      responses:
        '200':
          description: Model status
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ModelStatusResponse'
        '404':
          description: Model not found
  /v1/models/{model_name}/metadata:
    get:
      operationId: getModelMetadata
      summary: Google TensorFlow Get model metadata
      description: >-
        Returns metadata for a model including signature definitions that
        describe the model's inputs and outputs.
      tags:
        - Model Status
      parameters:
        - $ref: '#/components/parameters/model_name'
      responses:
        '200':
          description: Model metadata
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ModelMetadataResponse'
  /v1/models/{model_name}:predict:
    post:
      operationId: predict
      summary: Google TensorFlow Predict
      description: >-
        Runs prediction (inference) on a model. Accepts input tensors and
        returns output tensors from the model's serving signature.
      tags:
        - Prediction
      parameters:
        - $ref: '#/components/parameters/model_name'
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/PredictRequest'
      responses:
        '200':
          description: Prediction results
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/PredictResponse'
        '400':
          description: Invalid input
        '404':
          description: Model not found
  /v1/models/{model_name}:classify:
    post:
      operationId: classify
      summary: Google TensorFlow Classify
      description: >-
        Runs classification on a model. Returns class labels and scores
        for the given input examples.
      tags:
        - Prediction
      parameters:
        - $ref: '#/components/parameters/model_name'
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/ClassifyRequest'
      responses:
        '200':
          description: Classification results
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ClassifyResponse'
  /v1/models/{model_name}:regress:
    post:
      operationId: regress
      summary: Google TensorFlow Regress
      description: >-
        Runs regression on a model. Returns regression values for the
        given input examples.
      tags:
        - Prediction
      parameters:
        - $ref: '#/components/parameters/model_name'
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/RegressRequest'
      responses:
        '200':
          description: Regression results
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/RegressResponse'
components:
  parameters:
    model_name:
      name: model_name
      in: path
      required: true
      description: The name of the model
      schema:
        type: string
  schemas:
    PredictRequest:
      type: object
      properties:
        signature_name:
          type: string
          description: The serving signature to use (defaults to serving_default)
        instances:
          type: array
          description: Input instances for row-format requests
          items: {}
        inputs:
          description: Input tensors for columnar-format requests
          oneOf:
            - type: object
              additionalProperties: true
            - type: array
              items: {}
    PredictResponse:
      type: object
      properties:
        predictions:
          type: array
          description: Output predictions in row format
          items: {}
        outputs:
          description: Output tensors in columnar format
    ClassifyRequest:
      type: object
      properties:
        signature_name:
          type: string
        context:
          type: object
          additionalProperties: true
        examples:
          type: array
          items:
            type: object
            additionalProperties: true
    ClassifyResponse:
      type: object
      properties:
        results:
          type: array
          items:
            type: object
            properties:
              classes:
                type: array
                items:
                  type: object
                  properties:
                    label:
                      type: string
                    score:
                      type: number
    RegressRequest:
      type: object
      properties:
        signature_name:
          type: string
        context:
          type: object
          additionalProperties: true
        examples:
          type: array
          items:
            type: object
            additionalProperties: true
    RegressResponse:
      type: object
      properties:
        results:
          type: array
          items:
            type: object
            properties:
              value:
                type: number
    ModelStatusResponse:
      type: object
      properties:
        model_version_status:
          type: array
          items:
            type: object
            properties:
              version:
                type: string
              state:
                type: string
                enum: [START, LOADING, AVAILABLE, UNLOADING, END]
              status:
                type: object
                properties:
                  error_code:
                    type: string
                  error_message:
                    type: string
    ModelMetadataResponse:
      type: object
      properties:
        model_spec:
          type: object
          properties:
            name:
              type: string
            version:
              type: string
        metadata:
          type: object
          additionalProperties: true
TensorFlow Serving REST API

Documentation

Specifications

Schemas & Data

Other Resources

OpenAPI Specification
