Qwen, DeepSeek, GLM, Kimi and more open-weight models behind one OpenAI-compatible endpoint. Start with 500,000 free quota ($1) — up to 9M+ tokens — then pay only for what you use. No subscription.
curl https://api.novatoken.dev/v1/chat/completions \ -H "Authorization: Bearer sk-…your NovaToken key" \ -H "Content-Type: application/json" \ -d '{ "model": "deepseek-chat", "messages": [{"role": "user", "content": "Hello!"}] }'
Not sure which to pick? Set model: "auto" and NovaToken routes each request to the best-capable, cheapest model that fits — you're billed at that model's actual rate.
$5 one-off pack or monthly plans. No subscription lock-in — unused balance never expires.
One auto model picks the right engine per request, and you pay the real per-token rate of whichever model answers.
OpenAI protocol out of the box, plus an Anthropic-compatible endpoint for Claude Code. Works with LangChain, LlamaIndex, n8n, curl.
A live usage dashboard shows every call, every token, every cent — in your console, no support ticket needed.
One flat price, credits auto-refill on your billing date, cancel anytime.
Launch promo: first month 50% off with code PRODUCTHUNT · free tokens credited automatically at sign-up
* $1.00 starting credit, billed at real usage rates (70% input / 30% output mix): up to ≈9M+ tokens on DeepSeek (even more with cache discounts) · 7.5M on Qwen · 3.9M on GLM · 330k on Kimi. Unused credit never expires.
Sign up, claim 500,000 free quota ($1), create a key, change one line of base URL. That's the whole migration.
Create your API key →