General FAQ

What is Chuizi.AI?

Chuizi.AI is a unified AI API gateway. With one baseURL and one API key, you get access to 221 models across 18 primary providers including OpenAI, Anthropic, Google, DeepSeek, Qwen, GLM, Kimi, MiniMax, and more.

Requests use the OpenAI-compatible protocol by default, with native Anthropic and Gemini protocols also available. Your existing code, SDKs, and tools work instantly — just change your baseURL and API key.

How is this different from calling each provider directly?

Calling each provider directly means managing separate accounts, multiple API keys, different billing systems, and inconsistent error formats. Chuizi.AI solves all of that:

• Unified billing — one balance for all models
• Automatic failover — when a provider goes down, requests route to backup channels automatically
• Consistent error format — all errors return in OpenAI-compatible format, no provider-specific error handling needed
• One key for everything — a single ck- prefixed API key works across all models and protocols

What models are supported?

221 models across these providers:

• OpenAI — GPT-4.1, GPT-5, o3, o4-mini, DALL-E, and more
• Anthropic — Claude Opus 4-6, Claude Sonnet 4-6, Claude Haiku 4-5
• Google — Gemini 2.5 Pro, Gemini 2.5 Flash, Imagen 4.0
• DeepSeek — DeepSeek V3.2, DeepSeek R1
• Meta — Llama 4 series
• Mistral — Mistral Large, Codestral
• Chinese providers — Qwen 3.6, GLM-5.1, Kimi, Doubao, MiniMax

See the full list on the Models & Pricing page.

How is pricing calculated?

Upstream provider cost x 1.05. No hidden fees. For example:

• Upstream input price of $3.00/1M tokens becomes $3.15/1M tokens on Chuizi.AI
• Upstream output price of $15.00/1M tokens becomes $15.75/1M tokens on Chuizi.AI

All prices are listed transparently on the Models & Pricing page and updated in real time.

Is my data secure?

Yes. Chuizi.AI does not store any request or response message content. We only log metadata (model name, token counts, latency, status code) for billing and usage tracking.

Request data is encrypted via HTTPS, forwarded to the upstream provider for processing, and then discarded. API keys are securely encrypted before storage — the plaintext key is shown only once at creation time.

How do I access Chuizi.AI from China?

All models are accessible via api.chuizi.ai directly:

• Chinese models (Qwen, GLM, Kimi, Doubao, etc.) — routed through Asia-Pacific nodes for low latency
• International models (OpenAI, Anthropic, Google, etc.) — routed through the Chuizi.AI gateway. Users in China can access api.chuizi.ai directly

What API protocols are supported?

Three native protocols covering all major use cases:

All three protocols share the same authentication, billing, and rate limiting system.

Is there an SLA?

Chuizi.AI targets 99.9% uptime. The gateway supports automatic failover and circuit breaking — when an upstream provider experiences an outage, requests are automatically rerouted to backup channels to minimize disruption.

How do I get started?

Sign up at app.chuizi.ai using GitHub or Google. Create an API key, top up at least $5 (Stripe / Alipay / card), then start making requests. Pay-as-you-go from there at upstream cost × 1.05. Follow the First Request guide to make your first API call in under a minute.

Next Steps

• First Request — Make your first API call in under a minute
• Choose a Model — Pick the right model for your use case
• Billing Model — Understand how pricing and billing work