AI model pricing API

A free, read-only API for LLM pricing per million tokens, context windows, capabilities and hosting across AI providers. One consistent JSON shape, refreshed daily, no API key to try it.

Models
864
Providers and hosts
143
Last updated

Quick start

Search for a model by name:

curl "https://api.modelmonitor.dev/models"
{
  "total": 3,
  "data": [
    {
      "id": "anthropic/claude-sonnet-5-5",
      "name": "Claude Sonnet 5.5",
      "provider": "anthropic",
      "context_window": 1000000,
      "pricing": { "input_per_million": 2, "output_per_million": 10 },
      "capabilities": { "tool_calling": true, "reasoning": true, "vision": true },
      "hosted_on": [{ "provider": "Amazon Bedrock", "slug": "amazon-bedrock", "…": "…" }]
    }
  ]
}

Prices are per million tokens in USD. typical_request gives the cost of a 2,000 input and 500 output token call, so models compare at a glance.

Endpoints

Examples

Cheap models

Chat models costing at most $0.10 per million input tokens.

curl "https://api.modelmonitor.dev/models?type=chat&max_input_price=0.1"

One provider

Everything a company has created, using an id from /providers.

curl "https://api.modelmonitor.dev/models?provider=anthropic"

Where to run a model

Models offered by a host, such as Amazon Bedrock.

curl "https://api.modelmonitor.dev/models?hosted_on=amazon-bedrock"

Another currency

Prices converted from USD; the rate used comes back as exchange_rate.

curl "https://api.modelmonitor.dev/models?q=sonnet&currency=GBP"

Paging

Up to 200 models per request.

curl "https://api.modelmonitor.dev/models?limit=50&offset=50"

Need more?

The free API is rate limited. Paid access removes the limit and adds webhook notifications, so we call your endpoint whenever a model’s pricing or capabilities change. To buy access, email contact@modelmonitor.dev. We’re happy to talk through what you need.

AI model pricing FAQ

Do I need an API key?

No. Light use is free and keyless, with cached responses.

How fresh is the data?

Refreshed daily. Every response carries an updated_at timestamp; the latest is 2026-11-10.

What is the difference between provider and host?

A provider created the model. A host offers it, often at a different price, context window or uptime. Each model lists every host under hosted_on.

Can I be told when a price changes?

Yes. Paid plans also remove the rate limit and call your webhook whenever a model’s pricing or capabilities change.

Can I generate a client?

Yes. The API publishes an OpenAPI 3.1 spec at /openapi.json.