How to Use DeepSeek and Qwen API Outside China

#llm #ai #api #python

If you've been following the AI space lately, you've probably heard of DeepSeek and Qwen — two powerful Chinese large language models that are giving GPT-4 a serious run for its money. DeepSeek's reasoning capabilities are impressive, and Qwen handles multilingual tasks exceptionally well.

But here's the problem: accessing these APIs from outside China is a pain.

The Challenge

Chinese AI providers like DeepSeek, Qwen (Alibaba), Kimi (Moonshot), GLM (Zhipu), and Doubao (ByteDance) all have their own separate API endpoints, authentication systems, and billing structures — often documented only in Chinese. For developers outside China, this creates real friction:

Payment methods often require Chinese bank accounts
API formats differ between providers
Switching between models means rewriting integration code
Rate limits and reliability vary significantly

Enter ChinaWHAPI

ChinaWHAPI is an OpenAI-compatible API gateway for Chinese LLMs. Instead of juggling five different SDKs and accounts, you get one unified endpoint that works just like the OpenAI API you already know.

Supported models include:

DeepSeek (deepseek-chat, deepseek-reasoner)
Qwen (qwen-turbo, qwen-plus, qwen-max)
Kimi (moonshot-v1-8k, moonshot-v1-32k)
GLM (glm-4, glm-4-flash)
Doubao (doubao-pro-4k, doubao-pro-32k)

Quick API Example

Since ChinaWHAPI is OpenAI-compatible, you can use the official OpenAI Python SDK with zero code changes:

from openai import OpenAI

client = OpenAI(
    api_key="your-chinawhapi-key",
    base_url="https://api.chinawhapi.com/v1"
)

response = client.chat.completions.create(
    model="deepseek-chat",
    messages=[
        {"role": "user", "content": "Explain quantum computing in simple terms"}
    ]
)

print(response.choices[0].message.content)

Want to switch to Qwen? Just change the model name:

model="qwen-plus"

That's it. No SDK changes, no authentication rewiring, no new billing accounts.

Use Cases

Cost optimization: Chinese LLMs are significantly cheaper than GPT-4 for many tasks
Multilingual apps: Qwen and GLM excel at Chinese-language and mixed-language processing
Agent workflows: Works seamlessly with LangChain, LlamaIndex, Dify, and Cursor
Research & prototyping: Access cutting-edge Chinese AI models without the setup headache