# trustai.center pay

> Zero-friction crypto payment and inference layer for AI agents. OpenAI-compatible chat completions paid with stablecoins: prepaid credit keys (tai_) topped up in USDC/USDT, or per-request x402 payments in USDC on Base. No card, no account, no signup.

Base URL: https://pay.trustai.center/v1 (OpenAI-compatible)

## Endpoints
- GET  /v1/status — whether paid inference is currently available
- GET /v1/models — servable text chat models with USD prices (each has servable:true, context.max_output_tokens). ?include=discovery adds image/video models listed for reference only (servable:false). Query: modality, max_price (USD per 1M output tokens), sort=popularity|price|price_desc, q, limit. Text models are served via /v1/chat/completions.
- Pricing: refreshed daily; models sold at the model maker's list price (never more than 2% above). Each model has pricing.list_*_per_million_tokens and pricing.vs_list.
- GET /v1/trending — top models by tokens served in the last 24h (refreshed daily at 08:00 UTC).
- MCP list_models takes the same filters plus include="discovery".
- POST /v1/chat/completions — `Authorization: Bearer tai_...` (debits credits) or x402 (`X-PAYMENT` header, USDC on Base). Without either, returns HTTP 402 with x402 payment requirements. Supports `stream: true` (SSE).
- Reasoning models (glm-5.3-flash, gpt-oss-120b): max_tokens is raised to at least 1024; thinking is returned in message.reasoning_content. JSON bodies are accepted with or without Content-Type: application/json.
- POST /v1/topup {"pack":1|3|10|25 (1 = starter),"chain":"base"|"solana"|"bsc","promo"?} — returns deposit address, exact amount, and a new tai_ key if none was sent
- POST /v1/topup/verify {"order_id","tx_hash","signature"?} — credits the amount actually received. Exact amount = automatic; within -0.5% or $0.01 (or overpaid) = credited as received, promo pro rata; other amounts need "signature".
- POST /v1/topup/claim {"chain","tx_hash","signature"} — credit a stablecoin transfer to our address without a matching order. signature = sending wallet signs "trustai.center claim <chain>:<tx_hash>" (EVM personal_sign, Solana ed25519). Max 7 days old, each tx once; omit Bearer key to get a new key.
- GET  /v1/balance — balance, spend cap, spent (Bearer tai_ key)
- POST /v1/keys/cap {"spend_cap_usd": number|null} — set a spend cap on a key
- POST /v1/keys/limits {"spend_cap_usd"?, "daily_limit_usd"?, "per_request_max_usd"?} — spend controls (null removes, omitted = unchanged; daily resets 00:00 UTC)
- GET  /v1/usage?limit=50 — audit log for a key: charges, top-ups, last 24h totals (Bearer tai_ key)
- GET  /v1/stats — public usage stats from our database (tokens/requests 24h, keys created, top models 7d, promo redemptions). Real numbers only.
- GET  /v1/health and /status — live checks: inference upstream, Base/Solana/BSC RPC, x402 facilitator, last catalog sync
- Limits: max_tokens up to 16000 (or model max if lower; default 4096), 400000 input chars within model context, 500 messages. Balance and limits are checked against the worst-case cost before the model is called.
- Promo: "promo":"LAUNCH20" on /v1/topup = +20% credits on a key's first top-up, until 2026-11-01 23:59 Europe/Paris.
- MCP (streamable HTTP): https://pay.trustai.center/mcp — tools: list_models, list_trending, get_balance, get_usage, set_cap, create_topup, verify_topup, chat_completion, get_stats

## Payment rails
- Credits: USDC on Base, USDC on Solana, USDT or USDC on BNB Chain. Send the exact amount quoted by /v1/topup.
- x402: scheme "exact", network "base", asset USDC 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913, payTo 0x5beB45770465cBd7BDb8517b64fB52F6dD6607cf. Settlement only happens after a successful completion.

## Docs
- [Docs and quick start](https://pay.trustai.center/docs)
- [Terms](https://pay.trustai.center/conditions)
- [Agent manifest](https://pay.trustai.center/.well-known/trustai.json)

## Guides
- [Best LLM API for AI agents (2027)](https://pay.trustai.center/guides/best-llm-api-for-ai-agents-2027)
- [Pay for an LLM API with crypto / USDC](https://pay.trustai.center/guides/pay-for-llm-api-with-crypto-usdc)
- [x402 LLM API](https://pay.trustai.center/guides/x402-llm-api)
- [MCP server for inference](https://pay.trustai.center/guides/mcp-server-for-inference)
- [DeepSeek V4.1 Flash API price](https://pay.trustai.center/guides/cheapest-deepseek-v4-1-flash-api)
- [gpt-oss-120b API price](https://pay.trustai.center/guides/cheapest-gpt-oss-120b-api)
- [Qwen3 235B API price](https://pay.trustai.center/guides/cheapest-qwen3-235b-api)
- [Kimi K2 Thinking API price](https://pay.trustai.center/guides/cheapest-kimi-k2-thinking-api)