DashScope, Zhipu, DeepSeek: OpenAI-Compatible API Base URLs in China (2026)

TL;DR — Which Provider, in 30 Seconds

OpenAI-Compatible Base URLs and Endpoint Setup

Every provider follows the same pattern: set base_url, set api_key, pick a model. DashScope's official OpenAI-compatible endpoint is https://dashscope.aliyuncs.com/compatible-mode/v1; Zhipu's GLM endpoint is https://open.bigmodel.cn/api/paas/v4.

Code: One SDK, Six Providers

Switching providers is a one-line change. The same OpenAI Python SDK call works for all six — only base_url, api_key, and model differ.

from openai import OpenAI

# SiliconFlow — 200+ models on one key
client = OpenAI(
    base_url="https://api.siliconflow.cn/v1",
    api_key="your-siliconflow-key",
)
response = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-V3",
    messages=[{"role": "user", "content": "Explain transformer attention."}],
)

# DashScope — switch two lines for official Qwen
client = OpenAI(
    base_url="https://dashscope.aliyuncs.com/compatible-mode/v1",
    api_key="your-dashscope-key",
)
response = client.chat.completions.create(
    model="qwen3.7-plus",
    messages=[{"role": "user", "content": "Explain transformer attention."}],
)

# DeepSeek direct — lowest per-token cost
client = OpenAI(
    base_url="https://api.deepseek.com",
    api_key="your-deepseek-key",
)
response = client.chat.completions.create(
    model="deepseek-v4-flash",
    messages=[{"role": "user", "content": "Explain transformer attention."}],
)
# Same pattern for Volcengine, Zhipu, and MiniMax — only base_url + api_key + model change.

Pricing Comparison

Per million tokens in CNY (¥), sourced from official pricing pages, June 2026.

* Prices from official pricing pages on June 4–5, 2026. Verify before procurement — all providers adjust pricing frequently.

Rate Limits and Free Tiers

Decision Tree: Pick Your Provider

Route Through All Providers at Once

If you don't want to pick just one, a routing layer distributes requests across providers based on model, cost, and availability. TheRouter routes OpenAI-compatible requests through all six providers and handles fallback automatically — so you can use SiliconFlow as your primary for DeepSeek models and DashScope for Qwen, with automatic failover when a provider is unavailable.

# One TheRouter key → all six providers, automatic fallback
curl https://api.therouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $THE_ROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-v4-flash",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

One key. Six providers. Automatic routing. Read the docs →

FAQ