Ali releases multimodular intelligence modelQwen3.7-Plus

Ali Baba officially released the latest multi-mode model, Qwen 3.7-Plus, in the early hours of this morning, as an intelligent base that integrates visual and linguistic abilities and expands visual-linguistic capabilities across the board on the basis of Qwen3.7 text skills. In terms of performance, Ali claims that Qwen 3.7-Plus is among the top five in China on the list of global visual models. In terms of pure text capability, the model is strong on the coding smart body benchmark Terminal Bench 2.0, the SWE-bench series and SciCode, and is at the top of the Plus level model on the reasoning benchmark GPQA Diamond, HMMT and IMOANswerBench, which is close to the Max level model level as a whole. In terms of polymorphics, the model has significantly improved in terms of previous generation assessments of GUI intelligence such as ScreenSpot Pro, OSWorld-Verified and AndroidWorld, as well as in terms of video understanding missions and multiple driving scenes such as VideoMMU, MLVU。