DEV Community

xu
xu

Posted on

How to Access 49 Chinese AI Models with One API

How to Access 49 Chinese AI Models with One API

Chinese AI labs are shipping incredible models — DeepSeek V4, Kimi K3, Qwen 3.7, Doubao, GLM 5.2 — often at a fraction of the cost of Western alternatives. But for developers outside China, accessing them is a nightmare: separate API keys, CNY pricing, region blocks.

I built Express API to solve this: a single OpenAI-compatible endpoint for 49 Chinese models.

The Problem

To use Chinese AI models from overseas, you typically need:

  • A separate account on each platform (DeepSeek, Moonshot/Kimi, Alibaba Cloud, Volcengine, Zhipu...)
  • Chinese phone number verification for some
  • CNY payment methods
  • Different API formats for different vendors

The Solution

One API key, one endpoint, USD billing:

const client = new OpenAI({
  apiKey: "YOUR_API_KEY",
  baseURL: "https://gptapi.net.cn/v1"
});

// Works with ANY model from 10 vendors
client.chat.completions.create({
  model: "kimi-k3",  // or deepseek-v4-pro, qwen3.7-max...
  messages: [{ role: "user", content: "Hello!" }]
});
Enter fullscreen mode Exit fullscreen mode

Models Available (49)

Vendor Top Models
DeepSeek V4 Pro, V4 Flash
Moonshot/Kimi K3, K2.7 Code, K2.6, K2.5
Alibaba/Qwen 3.7 Max, 3.7 Plus, 3.6 Plus/Flash
ByteDance/Doubao Pro 128K, Pro 32K, Lite 128K
Zhipu/GLM 5.2, 5.1, 5, 4.7, 4.6, 4.5
MiniMax M3, M2.7
Baidu/ERNIE 5.1, 4.5 Turbo
Tencent/Hunyuan HY3
Xiaomi/MiMo V2.5, V2.5 Pro
StepFun 3.7 Flash, 3.5 Flash

Pricing

Transparent per-token billing with separate input/output rates:

  • DeepSeek V4 Flash: $0.16 / $0.32 per 1M tokens
  • Kimi K3: $3.25 / $16.23 per 1M tokens
  • Qwen 3.7 Plus: $0.32 / $1.30 per 1M tokens

Full pricing: gptapi.net.cn/pricing

Try It

Feedback welcome! What Chinese models would you like to see added?

Top comments (0)