API Reference

The OpenModels API provides programmatic access to the model registry, provider data, and real-time telemetry. All endpoints return JSON and follow consistent conventions for pagination and error handling.

Base URL

All API requests use the following base URL:

https://api.openmodels.run

All endpoints are prefixed with /api/. For example, to list models:

https://api.openmodels.run/api/models

Authentication

The OpenModels API is public and read-only. No authentication, API keys, or tokens are required. You can start making requests immediately.

# No auth headers needed
curl "https://api.openmodels.run/api/models"

Request Format

• Protocol: HTTPS only
• Methods: GET (all endpoints are read-only)
• Content-Type: Not required for GET requests
• Query parameters: Used for filtering, searching, and pagination

curl "https://api.openmodels.run/api/models?search=deepseek&capability=reasoning&limit=10"

Response Format

All responses return JSON with a consistent structure. List endpoints wrap results in a data array with pagination metadata:

{
  "data": [
    { "id": "deepseek-v3", "name": "DeepSeek V3", "..." : "..." }
  ],
  "pagination": {
    "page": 1,
    "limit": 20,
    "total": 44
  }
}

Single-resource endpoints return the object directly:

{
  "id": "deepseek-v3",
  "name": "DeepSeek V3",
  "vendor": "deepseek",
  "capabilities": ["chat", "reasoning", "code"],
  "context_window": 128000,
  "max_output_tokens": 8192
}

Pagination

All list endpoints support pagination via query parameters:

Example request:

curl "https://api.openmodels.run/api/models?page=2&limit=10"

Pagination metadata in response:

{
  "pagination": {
    "page": 2,
    "limit": 10,
    "total": 44
  }
}

To calculate total pages: Math.ceil(total / limit).

Error Format

When a request fails, the API returns a JSON error response with a consistent structure:

{
  "error": {
    "status": 404,
    "message": "Model not found",
    "type": "NOT_FOUND"
  }
}

Error Response Fields

Common Error Types

Rate Limiting

The API applies basic rate limiting to ensure fair usage:

• Limit: 100 requests per minute per IP address
• Headers: Rate limit status is returned in response headers

When rate limited, the API returns a 429 status with an error response:

{
  "error": {
    "status": 429,
    "message": "Rate limit exceeded. Try again in 45 seconds.",
    "type": "RATE_LIMITED"
  }
}

Available Endpoints

Quick Example

Fetch the first 5 models with vision capability:

curl "https://api.openmodels.run/api/models?capability=vision&limit=5"

{
  "data": [
    {
      "id": "gpt-4o",
      "name": "GPT-4o",
      "vendor": "openai",
      "capabilities": ["chat", "vision", "code", "reasoning"],
      "context_window": 128000,
      "max_output_tokens": 16384
    }
  ],
  "pagination": {
    "page": 1,
    "limit": 5,
    "total": 12
  }
}