Getting started

The ModelMonitor API returns pricing, capabilities and hosting for AI models. It is read-only, and free use needs no API key.

Main features

• One normalised schema. Pricing, capabilities and hosting for AI models in a single consistent JSON shape, whichever provider made or hosts the model.
• Fresh data. Refreshed daily, so prices and capabilities stay current.
• Search and filter. Narrow by provider, host, type, name and maximum input price.
• Multiple currencies. Prices are in USD, or converted on request with the rate returned.
• Free to try. No key needed for light use, with cached responses.
• Webhook events. Paid plans get a call to your endpoint whenever a model’s attributes change.
• OpenAPI spec. Generate a client in any language.

Make a request

curl "https://api.modelmonitor.dev/models"

{
  "total": 3,
  "updated_at": "2026-10-29T14:46:28.871Z",
  "version": 1,
  "data": [
    {
      "id": "anthropic/claude-sonnet-5-5",
      "name": "Claude Sonnet 5.5",
      "provider": "anthropic",
      "type": "chat",
      "description": "Claude Sonnet 5.5 is Anthropic's Sonnet-class model for well-scoped everyday work, succeeding Claude Sonnet 5 as a direct upgrade. It is especially strong at building features, fixing bugs, and producing...",
      "context_window": 1000000,
      "max_output_tokens": 128000,
      "pricing": {
        "input_per_million": 2,
        "output_per_million": 10,
        "typical_request": { "cost": 0.009, "input_tokens": 2000, "output_tokens": 500 }
      },
      "capabilities": { "tool_calling": true, "reasoning": true, "vision": true },
      "intelligence_score": 54.2,
      "hosted_on": [
        {
          "provider": "Amazon Bedrock",
          "slug": "amazon-bedrock",
          "offerings": [
            {
              "tier": null,
              "quantization": null,
              "pricing": {
                "input_per_million": 2,
                "output_per_million": 10,
                "typical_request": { "cost": 0.009, "input_tokens": 2000, "output_tokens": 500 }
              },
              "price_difference_percent": 0,
              "context_window": 1000000,
              "uptime_percent": 100
            }
          ]
        }
      ]
    }
  ],
  "limit": 1,
  "offset": 0,
  "currency": "USD",
  "exchange_rate": null
}

Filtering and paging

Combine provider, hosted_on, type, q and max_input_price to narrow results.

• q is a search: every space-separated term must appear in the model’s id, name or provider, so q=anthropic sonnet works.
• provider is the company that created the model; hosted_on is a company that offers it, such as amazon-bedrock.
• Both take a provider id. GET /providers lists every valid id with the number of models it created and hosts. Page with limit (1 to 200, default 50) and offset.

Currencies

Prices are in USD. Pass currency=GBP to convert them; the allowed codes are listed under the currency parameter of List models; the rate used is returned in exchange_rate.

Authentication

Free use needs no authentication. Paid access removes the rate limit and adds webhook events: ModelMonitor calls your endpoint whenever a model’s attributes change, such as its pricing or capabilities.

Paid access is priced per customer. To get a quote and an API key, email contact@modelmonitor.dev describing your expected usage, such as requests per day and whether you need webhook events. Send the key in an X-API-Key header:

curl -H "X-API-Key: YOUR_API_KEY" "https://api.modelmonitor.dev/models?q=claude"

To receive webhook events, include the URL of your endpoint in the same email.

Rate limits

Free use is limited to 5 requests per minute. Past that, GET /models and GET /providers return 429 with a Retry-After header. Responses are cached for a minute, so repeating an identical request is usually served from cache and does not count against the limit. Paid plans send their key in an X-API-Key header and are not rate limited.

Errors

Invalid query parameters return 400 with a body of the form { "error": "message" }.

OpenAPI

The machine-readable spec is at /openapi.json.

Next steps