What is ZeroFA?
ZeroFA is an OpenAI-compatible multi-model API gateway. One sk-zerofa-* key can call Claude, GPT, Gemini, Qwen, DeepSeek, and more, with real-time usage charges deducted from a configured currency wallet.
How is this different from using OpenAI directly?
The API protocol stays compatible, so existing OpenAI SDK code needs only base_url and api_key changes. ZeroFA lets one key switch across providers while centralizing billing, usage, and rate limits.
Which models are supported?
Use the live Model Catalog as the source of truth, or retrieve the list with GET /v1/models.
How does billing work, and can I be overcharged?
ZeroFA uses a prepaid wallet with real-time usage billing, and failed requests are not charged. Each key can have a total budget, RPM limit, and daily request limit. See Pricing and billing
Is the API wallet the same as a web chat membership?
No. They are completely separate. The API wallet is charged only when an sk-zerofa-* key calls the API; web chat subscriptions and credits use a different balance.
What should I do if a key is exposed?
Immediately open API Keys to revoke the key and create a replacement. A secret is shown in full only once. Store it securely, never place it in frontend code, and do not commit it to a repository.
Are streaming responses supported?
Yes. Set stream: true on chat/completions to receive SSE chunks, ending with data: [DONE].
Can I use the native Anthropic or Gemini SDK?
Yes. The OpenAI-compatible endpoint automatically translates function calling and multimodal features. To reuse an official SDK directly, use the native Anthropic Messages or Gemini generateContent endpoint. Authentication, billing, and routing still run through ZeroFA after you change base_url. Anthropic Messages / Gemini generateContent
How can I test without writing code?
Send one of the curl or SDK examples in the docs, or use the request runner at the top of an endpoint page.