Developer Docs › FAQ

FAQ

Frequently asked questions about integration, billing and choosing models.

Integration

Should the base URL include /v1?

  • "OpenAI-compatible" settings such as the OpenAI SDK, LangChain (OpenAI), Cursor and Dify: include /v1, i.e. https://api.onefast.ai/v1.
  • Anthropic SDK, Claude Code, Gemini SDK / CLI and Cherry Studio: don't include it; use https://api.onefast.ai.

A 404 almost always means this is set wrong.

Can I call the API directly from a web front end?

Technically yes, but it exposes your key to every visitor. Route requests through your own back end and keep the key on the server.

What timeout should I use?

  • Regular chat: 60–120 seconds.
  • Reasoning models, long outputs, Claude Code: 600 seconds.
  • Image generation: 300 seconds. Video is an async task — the submit call returns quickly, so no long timeout is needed.

Models & groups

How do I see which models are available and what they cost?

See the model catalog, which updates in real time. When you're logged in, it shows every group available to your account and your actual prices. You can also call GET /v1/models with your key to list the models available to that key.

What's the difference between the same model in different groups?

Price, stability and concurrency. For production workloads that need high stability, pick "premium / official" groups; for batch or offline jobs, "discount" groups work well. See Groups & billing.

Can I use several groups at once?

Yes. A key can have several groups in a set order; if the first group doesn't have the model, the request falls through to the next one. You can also pin a group with the X-Token-Group header. See Authentication · Choosing a group.

Billing

Are failed requests charged?

No. Requests that return an error are not charged, and failed video tasks are refunded their pre-charge automatically.

Why was I charged more than I estimated?

Common reasons: reasoning / thinking tokens count as output; multi-turn conversations resend the entire history every time; image inputs are converted into tokens. Every charge is itemized on the "Logs" page in the console. Prompt caching can significantly cut the cost of long prefixes.

How do I check my balance?

It's shown on the console "Dashboard". You can also call GET /v1/dashboard/billing/subscription with your key; see Authentication · Checking your balance.

Data & security

What if my key leaks?

Immediately delete the key on the "API Keys" page in the console, create a new one, then check the logs for unusual calls. Data handling is governed by the site's privacy policy.