Frequently asked questions

What people worry about before their first API relay top-up.

Never used an API relay and nervous about paying: what is the smallest top-up I can try?

The smallest top-up is ¥10; the options are ¥10, ¥50, ¥100, ¥200, ¥500, ¥1000, paid by WeChat Pay or bank card in your dashboard. Accounts registered in this language get no signup credit. Your balance does not expire, and unused balance can be refunded within 7 days of the top-up, minus what was used and payment-channel fees: leave a message on the Contact page and it is handled within 3 business days.

Is 59API run by Anthropic or OpenAI, and what does an API relay actually do?

No, 59API is an independent third-party AI API relay, not owned by or partnered with Anthropic, OpenAI or Google. One prepaid account and one API key reach 26 public models, including Claude, GPT, Gemini and DeepSeek, through api.59api.com, and requests get to the models through upstream providers.

Can I use Claude Code or Codex without a Claude Pro or ChatGPT Plus subscription?

Yes. 59API needs neither subscription: calls are paid from your 59API balance and there is no monthly fee. Install Claude Code (2.1.285 or newer) or Codex, then swap in 59API's address and your key.

I have never set up an API key: what do I change for Claude Code, Codex and Gemini CLI?

Create a key in your dashboard, then change two settings per tool. Claude Code: ANTHROPIC_BASE_URL=https://api.59api.com and ANTHROPIC_AUTH_TOKEN=your key. Codex: base_url https://api.59api.com/v1 and api_key your key. Gemini CLI: GOOGLE_GEMINI_BASE_URL=https://api.59api.com and GEMINI_API_KEY=your key.

Can I paste my 59API key into Cursor and use Claude there?

No. Claude models run only in the official Claude Code or Claude Desktop client, version 2.1.285 or newer. Codex, Cursor, Cherry Studio, Cline, OpenCode and other OpenAI-compatible clients and SDKs work with GPT, DeepSeek and the remaining models, while Gemini models use Google's native protocol, for example through Gemini CLI, and are not on /v1/chat/completions.

Pay-as-you-go and tokens confuse me: what does one question to Claude Sonnet 5.5 cost, and could my bill suddenly jump?

About $0.005 (¥0.036), counting 2,000 input and 500 output tokens at Claude Sonnet 5.5's $1.11 input and $5.56 output per million tokens. Claude Code sends project context along, so real requests usually use more. Charges come only from the balance you prepaid, with input rates from $0.1389, separate cache rates, no monthly fee and no automatic payment; when the balance runs out requests return HTTP 402 until you top up. Billing is in CNY, and each request's tokens and cost appear in your dashboard.

I accidentally pushed my API key to GitHub or pasted it in a group chat: can someone spend my balance?

Yes, so act straight away: anyone holding the key can call models and spend your balance. Revoke that key in your dashboard and create a new one. A key works as an account credential, and losses from leaking or sharing it are the account holder's responsibility.

Do beginners need a monthly card, and can I keep going when its quota runs out?

No, pay-as-you-go works without a card. Monthly and weekly cards are optional prepaid quota, from $40.14, spent through their own key: a Claude card key calls only Claude models and a GPT card key only GPT models. When the quota or weekly cap is used up, that key returns HTTP 402 and your balance is not touched, so you continue with a balance key; a balance key does not draw on a card. Unused quota does not roll over.