# Models: AI model price reference (models.agentexchange.work)

> Per-token prices, context length, modalities and supported parameters for about 470 hosted AI models (every major lab and open-weight model), rebased to USD per million tokens, with a cheapest-fit ranking and side-by-side comparison. Source: OpenRouter's public model list, cached one hour. Pay per call over x402 (USDC on Base) through the Agent Exchange store gateway; no account or key. Free sample: https://models.agentexchange.work/sample.json

## Paid routes (x402, USDC on Base, through the Agent Exchange store gateway; no account or key)
- GET https://store.agentexchange.work/gw/models/models (0.002 USD per call): The model list: id, name, provider, prompt and completion price in USD per million tokens, context length, input and output modalities, reasoning flag; filters provider= (anthropic, openai, google, me
- GET https://store.agentexchange.work/gw/models/model (0.002 USD per call): One model in full: every price component (prompt, completion, cache read and write, web search, images, requests), context length, max completion tokens, modalities, supported parameters, knowledge cu
- GET https://store.agentexchange.work/gw/models/cheapest (0.002 USD per call): The cheapest models that fit a job: rank by blended price (3 parts prompt, 1 part completion) under constraints min_context=, modality= (image, file, audio), reasoning=true, provider=, exclude_free=tr

## Free
- https://models.agentexchange.work/
- https://models.agentexchange.work/sample.json
- https://models.agentexchange.work/.well-known/x402
- Catalogue of every route on the store: https://store.agentexchange.work/.well-known/x402
- MCP: https://store.agentexchange.work/mcp