# Model Access Gateway

> OpenAI-compatible chat completions for 100+ AI models (DeepSeek, GPT, Claude, Gemini, Grok, GLM, Qwen, Kimi, Llama, Mistral), paid per request in USDC on Base with x402. No account, no API key.

## How to call
1. POST https://agent402-arb.rosariovitale0096.workers.dev/v1/chat/completions with a JSON body: {"model": "<id from GET /models>", "messages": [{"role": "user", "content": "..."}], "max_tokens": 256}
2. The unpaid request returns HTTP 402. The PAYMENT-REQUIRED header holds an x402 v2 quote: scheme "exact", network eip155:8453 (Base), asset USDC 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913. The amount is the exact price for that prompt and max_tokens.
3. Sign the payment with any x402 client (EIP-3009 transferWithAuthorization) and resend the same request, with the same body, in the PAYMENT-SIGNATURE header.
4. The response is a standard chat.completion JSON. If the model fails to answer, the payment is refunded automatically.

## Endpoints
- POST https://agent402-arb.rosariovitale0096.workers.dev/v1/chat/completions: any model, chosen in body.model
- POST https://agent402-arb.rosariovitale0096.workers.dev/v1/chat/completions/{url-encoded model id}: one route per model (for example /v1/chat/completions/deepseek%2Fdeepseek-chat)
- GET https://agent402-arb.rosariovitale0096.workers.dev/models: live model list, free
- GET https://agent402-arb.rosariovitale0096.workers.dev/openapi.json: OpenAPI 3.1 with x-payment-info
- GET https://agent402-arb.rosariovitale0096.workers.dev/.well-known/x402: x402 discovery manifest
- GET https://agent402-arb.rosariovitale0096.workers.dev/terms: terms of use (paying for a request means accepting them)

## Limits
- max_tokens (or max_completion_tokens) from 1 to 16384; up to 256 messages; up to 200,000 characters per message; request body up to 384 KB; no streaming (stream must be false).
- Function calling (OpenAI format): tools, tool_choice, parallel_tool_calls, assistant tool_calls and role "tool" results. Message content is a string or an array of text parts; images are not accepted.
- Prices start at $0.0010 per request and depend on the model, prompt length and max_tokens. The 402 quote is always the final price.

## Short routes (one per model)
Same request and payment flow as /v1/chat/completions; the model is fixed by the path, so body.model can be omitted. GET on a route returns its 402 quote for a tiny fixed request. Docs page per model: https://agent402-arb.rosariovitale0096.workers.dev/docs/llm/<slug>.
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/claude-opus-5.5: Claude Opus 5.5
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/claude-sonnet-5.5: Claude Sonnet 5.5
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/claude-haiku-5.5: Claude Haiku 5.5
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/gpt-6-astra: GPT-6 Astra
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/gpt-6.1-sol: GPT-6.1 Sol
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/gpt-5.5: GPT-5.5
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/gpt-6-luna: GPT-6 Luna
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/gemini-4-argon: Gemini 4 Argon
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/gemini-3.1-pro: Gemini 3.1 Pro
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/gemini-3.8-flash: Gemini 3.8 Flash
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/grok-4.7: Grok 4.7
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/kimi-k3: Kimi K3
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/glm-5.3: GLM-5.3
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/deepseek-v4-pro: DeepSeek V4 Pro
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/deepseek-v4.1-flash: DeepSeek V4.1 Flash
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/deepseek-chat: DeepSeek V4 Flash Chat
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/qwen3.8-max: Qwen3.8 Max
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/minimax-m3: MiniMax M3
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/llama-3.3-70b: Llama 3.3 70B
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/mistral-large-4: Mistral Large 4
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/gpt-oss-120b: OpenAI GPT OSS 120B
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/claude-opus-5: Claude Opus 5
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/claude-sonnet-5: Claude Sonnet 5
- POST https://agent402-arb.rosariovitale0096.workers.dev/llm/gemini-3.6-flash: Gemini 3.6 Flash