# x402-apis

> Pay-per-request AI chat API. No accounts, no API keys. Pay as you go with
> USDC on Base via x402 (ERC-402) — HTTP 402 is the checkout.


- Home: https://x402-apis.archlinuxonthinkpad.workers.dev
- Payment protocol: x402, "exact" scheme, $0.001 USDC per request on eip155:8453
  (Base mainnet USDC 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913) to 0x25a47702b2852D82F4cc6DF13a7343213d9a2ECa. No subscription. Settled by
  facilitator; payer needs only USDC (gas is sponsored).
- How an agent pays: send a normal request; if a 402 is returned with a
  PAYMENT-REQUIRED header, sign and retry using an x402 client
  (e.g. @x402/fetch + @x402/evm/exact/client, @x402 Python, or any x402 wallet).
  Unpaid = 402. Paid = 200 with an OpenAI-style chat completion.
- Models: `@cf/meta/llama-3.1-8b-instruct-fp8`, `@cf/meta/llama-3.2-3b-instruct`, `@cf/meta/llama-3.3-70b-instruct-fp8-fast`, `@cf/qwen/qwen3.8-27b` (default `llama3.1-8b`).

## Endpoints

- GET / — service manifest (machine-readable index)
- GET /models — available models and price
- GET /stats — usage and settled revenue
- POST /api/chat — PAID. Body: {"model": optional, "messages": [{"role","content"}], "stream": false};
  returns OpenAI-style { id, choices:[{message:{content}}], usage }.
- GET /llms-full.txt — full plain-text docs for this service.
- GET /llms.txt — this index.

## Request shape (POST /api/chat)

```json
{ "model": "llama3.1-8b", "messages": [{ "role": "user", "content": "hi" }] }
```

## Error codes

- 402 Payment required — retry with an x402 payment signature.
- 400 invalid_json / messages_required / invalid_message
- 501 streaming_not_supported (stream: true not enabled)
- 502 ai_error

## Pricing

- Every POST /api/chat costs $0.001 in USDC on eip155:8453. One request = one payment.
- There is no API key, no signup, no subscription. Pay only when you call.