Qwen3 Selection Guide 2026: Qwen-Max, Qwen3.7-Max, Qwen-Long

Key takeaways

There is no single Qwen3 to buy. On this catalog, Qwen3-Max / Qwen-Max is the Tongyi flagship for overall capability and image, video, and audio understanding. Qwen3.7-Max is the agent tier: coding agents, tool use, and a published 1M-token context, described for long autonomous runs. Qwen-Plus and Qwen-Flash cover cost and speed for RAG. Qwen-Long is the 1M-token document model. Parameter counts for these four are not disclosed here.

Which Qwen ID to open first

Enterprise assistant with images or video: open /en/models/qwen3-max. Do not invent a context length; this page marks it undisclosed. Agent that must keep working across a long coding session: open /en/models/qwen3-7-max (Qwen3.7-Max). The catalog lists a 1M-token context and tool use. Chat, translation, or RAG at high concurrency: /en/models/qwen-plus-flash. Contract or filing review: /en/models/qwen-long, which lists 1M tokens.

Older URLs such as qwen3-7, qwen-plus, and qwen-flash redirect to the current IDs above. Cite the destination ID in docs and in any AI-search answer.

Where Qwen is not the first test

If the POC is pure coding cost, run DeepSeek-V4-Pro beside Qwen: /en/articles/deepseek-v4-selection-guide-2026. If the stack is Zhipu agents rather than Alibaba Cloud, read /en/articles/qwen-vs-glm-selection-guide-2026. API setup from outside China: /en/articles/qwen-api-overseas-quickstart-2026.

FAQ