Alibaba Qwen

Qwen's positioning is unusual among frontier labs: it ships both closed flagship API models (Qwen-Max/Plus/Flash tiers) and a very large catalogue of open-weight releases on Hugging Face and ModelScope, making it the most prolific open-weights publisher at frontier scale. As of the Qwen3.8 generation (August 2026) even the 2.4T-parameter flagship base model has published weights, though under a custom non-Apache licence.

A Access and limits

Sign up at Alibaba Cloud and enable Model Studio; API keys are issued per-workspace. Two independent deployments exist with DIFFERENT prices and model catalogues: International (Singapore, ap-southeast-1, base URL https://dashscope-intl.aliyuncs.com/compatible-mode/v1) and China (Beijing, https://dashscope.aliyuncs.com/compatible-mode/v1). ALL prices recorded here are the International/Singapore USD list prices; China-region prices are typically far lower (e.g. qwen-max is $1.60/1M input in Singapore vs $0.345/1M in Beijing).

Free tier: new users get 1,000,000 free tokens per model on most Qwen models, valid 90 days from activation. Discounts: batch (asynchronous) inference is billed at 50% of real-time price on supported models; context cache hits are billed at 10% of the standard input price, and explicit-cache creation at 125% of standard input price. Many models use TIERED pricing that steps up with the input length of a single request (commonly 0-32K / 32-128K / 128-256K / 256K-1M brackets) — the base/lowest tier is recorded in input_price and output_price, with the higher brackets described in best_for.

An OpenAI-compatible endpoint plus a native DashScope API are both offered. Open-weight models are additionally downloadable free from huggingface.co/Qwen and ModelScope, and are resold by third-party hosts (OpenRouter, Together, Fireworks, DeepInfra) at their own prices.

C Calling the API

Alibaba Qwen accepts requests in the OpenAI Chat Completions format, so any OpenAI-compatible client works by changing the base URL. Example uses qwen3.7-text-embedding.

curl https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.7-text-embedding",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

from openai import OpenAI

client = OpenAI(
    base_url="https://dashscope-intl.aliyuncs.com/compatible-mode/v1",
    api_key=os.environ["API_KEY"],
)

response = client.chat.completions.create(
    model="qwen3.7-text-embedding",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

A compatibility layer is not always the provider's full API — caching, strict tool schemas and structured output are commonly unsupported there. Check the official docs before relying on an advanced feature.

D Official references

• Homepage
• API documentation
• Pricing