Alibaba's Qwen team quietly launched the first previews of the Qwen3.7 series—Max and Plus—on chat.qwen.ai. The new version forces deep thinking mode and temporarily disables web search and code interpreter.
In the latest Arena rankings, Qwen3.7 claimed first place among Chinese models in both text and vision categories.
The flagship Qwen3.7-Max-Preview ranked 13th globally in the overall text leaderboard, pushing Alibaba's text R&D to 6th worldwide. The model cracked the global top 10 in hardcore reasoning tasks: 7th in math, 9th in expert prompting and software IT, and 10th in code generation.
Vision-focused Qwen3.7-Plus-Preview landed 16th globally in the vision rankings, boosting Alibaba's overall vision R&D to 5th place.
The previous flagship, Qwen3.6-Max-Preview, was released just 28 days ago on April 20. This rapid iteration—a preview arriving less than a month later—appears to be a pre-emptive launch to build momentum for the 2026 Alibaba Cloud Summit, kicking off May 20 in Hangzhou. At the summit, the company will unveil the technical underpinnings and commercial rollout of the new base model.
