321AI Model MarketplaceLive marketplace →

321AI FAQ (English)

The 12 questions developers ask most — answered with verifiable facts · data date 2026-11-20

Short version: multi-supplier marketplace, 165+ models, OpenAI + Anthropic protocol compatible, merchant-set prices, failures never billed, public price API at https://321ai.cc/api/public/model-prices.

What is 321AI?

321AI (321ai.cc) is a multi-supplier AI model marketplace: one account, one API endpoint, and 165+ models quoted live by many independent suppliers with public prices and upstream self-check records. Merchant-set prices.

How much cheaper is it than official pricing?

Market floor prices vary by model and move in realtime; the full table with supplier counts is public at /en/models/ and via the JSON API /api/public/model-prices. Judge with the live numbers rather than a static claim.

Which protocols are supported?

OpenAI protocol at base_url https://321ai.cc/v1 (chat completions and responses) and Anthropic protocol at https://321ai.cc (/v1/messages) — Claude Code works with one environment variable.

Is there a platform fee or commission?

No. Zero listing fees, zero commission, zero routing fee — 100% of your balance pays for model calls at quoted prices.

What happens when a request fails?

Failed requests are never billed. Pre-deducted quota is refunded automatically; only successful calls settle.

How do I pay?

Prepay with USDT (TRON/BSC/Polygon/Solana) or other channels shown in the console. Balance settles per call at actual usage.

Is there a free trial?

Yes — new accounts get trial credit you can spend on any model before topping up. Sign up at the console and create a key.

Are the models genuine?

Every listing names its merchant and channel, and merchants publish their own upstream self-check samples so you can pick the supplier you trust and pin it explicitly.

What are the route tiers (default/cheap/stable/fast/guaranteed)?

Per-key routing tiers: default is smart routing, cheap prioritizes price, stable prioritizes stability, fast prioritizes latency, guaranteed runs on merchant-supplied lanes. Pick per token; no code changes.

Can I use it with Cursor / Claude Code / Cherry Studio / Dify?

Yes — step-by-step guides for 20+ tools are under /en/guides/ (English) and /guides (Chinese), each with the exact fields to fill.

Do you store my conversations?

No. Request and response bodies are relayed in realtime only — the platform does not store or reuse them. Usage metadata (tokens, latency, status) is kept for billing and public quality stats.

Is there an API for prices?

Yes — GET /api/public/model-prices (JSON, CORS-open, no auth) returns input/output/cache prices, supplier count and a stability label for every model; every model page also has a machine-readable .md twin.

Still stuck?

Per-tool setup guides: https://321ai.cc/en/guides/ · per-model pricing: https://321ai.cc/en/models/.

All model pricesFeatured · 321AI communityFeatured · Open-source toolsFeatured · Advertise with usFeatured