Qwen API

The Qwen API is how you call Alibaba's Qwen models without hosting them: Alibaba Cloud Model Studio (formerly DashScope, and Bailian in mainland China) serves the closed Max, Plus and Flash models, vision, coding and omni models, hosted versions of the open-weight Qwen releases, and third-party models such as DeepSeek, Kimi, GLM and MiniMax. The current Max, Plus and Flash models take up to 1M tokens of context. For running the open-weight models yourself, see Qwen.

You can use the OpenAI SDK by pointing it at the /compatible-mode/v1 path, the Anthropic SDK or Claude Code at /apps/anthropic, or the native DashScope API and its dashscope SDK. Each region has its own endpoint, API key and model list, and a key only works in the region where it was created. Alibaba recommends a per-workspace domain ({WorkspaceId}.{region}.maas.aliyuncs.com); the older dashscope-intl.aliyuncs.com style domains keep working but get no new features after September 30, 2026.

Useful for a small team: one key covers Qwen plus several other model families; the Singapore region gives new accounts a free quota of about 1M tokens per model for 90 days, with a "Free Quota Only" switch that stops calls instead of billing once it runs out; batch calls cost half the real-time price; and explicit context-cache hits are billed at a fraction of normal input.

The main thing to choose is the region. Mainland China (Beijing), International (Singapore), Hong Kong, Tokyo, Frankfurt with an EU scope and Virginia with a US scope are offered; request data is stored in the region you call, and inference runs inside the deployment scope you pick. Batch inference and fine-tuning are only in Singapore and Beijing. Prices differ by region and are tiered by prompt length on some models.