Multi-merchant quotes, upstream self-checks, failures never billed
Every claim on this page can be re-verified. Prices and supplier counts come from our marketplace cache; the same data is published via free CORS-open APIs (https://321ai.cc/api/public/model-prices, https://321ai.cc/api/public/platform-stats) and a machine-readable .md twin for every model page. Data date 2026-11-20.
No listing fees, no matchmaking commission, no transaction cut, no routing fee. 100% of your balance pays for model calls at the quoted prices.
Market floor price and supplier count are published for every model — and exposed as JSON so you can audit them.
GET https://321ai.cc/api/public/model-prices
GET https://321ai.cc/api/public/platform-stats
Failed requests cost nothing; prepayments are refunded automatically. Only successful calls are billed.
On upstream 429/5xx the gateway retries and fails over automatically. Persistently bad channels enter a 1-hour cooldown, re-probed every 3 minutes, and return only after 2 consecutive successful probes. Live prices and supply: marketplace.
Every listing names its merchant and channel. Merchants publish their own upstream self-check samples, and you can pin the exact supplier you trust — the bill then follows that merchant's confirmed price.
171 models × 692 suppliers with 15,375 live quotes — every number re-fetchable from the public stats API.
Per-token (or per-call) billing at quoted prices, itemized in your usage logs. Top-ups via USDT (TRON/BSC/Polygon/Solana).
In-product ticket center with ~2-4h first response; billing and usage are itemized per request.
All figures on this page: 2026-11-20. Historical changes are visible in the marketplace UI; the stats API reports the current snapshot.