Refuel
Refuel Models API

Refuel hosts and finetunes LLMs purpose-built for data labeling. Tasks can use Refuel-hosted base models or custom finetuned models, with finetuning runs tracked and finetuned models reusable across applications.
Documentation GitHub OpenAPI
OpenAPI Specification

openapi: 3.0.1
info:
  title: Refuel Cloud API
  description: >-
    REST API for the Refuel Cloud data-labeling and data-enrichment platform.
    Refuel deploys a Task as a versioned Application served behind a REST
    endpoint; callers submit rows of structured / semi-structured data to the
    application's label endpoint to transform (label, clean, enrich) them in
    realtime using the application's LLM guidelines. Authentication uses an API
    key passed as a Bearer token. The only endpoint documented with a concrete
    request shape in Refuel's public documentation is the application label
    endpoint; dataset, task, application-management, and model/finetuning
    operations are performed in the Refuel app and via the Python and JavaScript
    SDKs.
  termsOfService: https://www.refuel.ai
  contact:
    name: Refuel
    url: https://www.refuel.ai
  version: '1.0'
servers:
  - url: https://cloud-api.refuel.ai
    description: Refuel Cloud API
paths:
  /applications/{applicationName}/label:
    post:
      operationId: labelWithApplication
      tags:
        - Applications
      summary: Transform rows with a deployed application
      description: >-
        Submit a JSON array of input rows to a deployed application to run LLM
        labeling / prediction in realtime. Each row is an object whose keys
        match the application's expected input fields. The application applies
        its Task guidelines and returns the transformed (labeled / enriched)
        output for each row. Subject to a documented rate limit of 300 requests
        per minute and up to 100 concurrent requests.
      parameters:
        - name: applicationName
          in: path
          required: true
          description: The name of the deployed application (for example, resume-parsing).
          schema:
            type: string
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/LabelRequest'
            example:
              - resume_link: https://path/to/resume.pdf
      responses:
        '200':
          description: Rows transformed successfully.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/LabelResponse'
        '401':
          description: Missing or invalid API key.
        '429':
          description: >-
            Rate limit exceeded (more than 300 requests per minute or more than
            100 concurrent requests).
      security:
        - BearerAuth: []
components:
  securitySchemes:
    BearerAuth:
      type: http
      scheme: bearer
      description: >-
        Refuel API key supplied as a Bearer token in the Authorization header:
        Authorization: Bearer <YOUR_API_KEY>.
  schemas:
    LabelRequest:
      type: array
      description: >-
        Array of input rows to transform. Each row is an object whose keys are
        the input fields expected by the target application.
      items:
        type: object
        additionalProperties: true
    LabelResponse:
      type: array
      description: >-
        Array of transformed rows. Each item contains the application's output
        fields; outputs may include confidence scores and explanations per field.
      items:
        type: object
        additionalProperties: true
security:
  - BearerAuth: []
Refuel Models API

Documentation

Specifications

Other Resources

OpenAPI Specification
