FAQ

Is this OpenAI?

No. The API shape is OpenAI-compatible. Chat models run on Gonka. Jev is not a Gonka model.

Why is it so cheap vs OpenRouter?

You pay the network’s token rate (governance × GNK/USD), not a lab’s list price. Input and output are the same $/M. OpenRouter charges input and output separately, and output is usually several times input.

Do I need GNK to use the API?

No. Register, use the free 1M tokens, then top up with card or crypto. GNK is an x2 bonus: GNK and WGNK buy twice the tokens of card or other crypto. That bonus is on the balance, not on the dashboard Rate / 1M. See Top up.

Which model should I use?

• DeepSeek V4 Flash — on Gonka. Reasoning, tools, long context
• MiniMax M2.7 — on Gonka. Throughput / agents
• GLM 5.3 Flash — on Gonka. Backup when the others are busy
• Jev — not a Gonka model. Free only after 100,000,000 tokens in the last 30 days. Otherwise there is no access. See Jev.

Always read /v1/models at runtime. See Available models.

Why did my model disappear?

Zero healthy shards, or a deprecated id. See Deprecated models and Status.

Are there rate limits?

No proxy RPM cap. Hosts cap concurrency. See Rate limits.

Can I use it from Cursor, Copilot, or Windsurf?

Cursor and GitHub Copilot chat or CLI, yes. Windsurf Cascade cannot take this base URL; use Cline or Roo Code inside Windsurf. See Agents and IDEs.

Do you train on my prompts?

By default we do not store prompt/response text. See Usage and Settings.

How do I white-label this?

Partners — child accounts, parent billing.

Who do I ask?

Telegram: t.me/gonka_gg.