Alibaba Cloud's Qwen large model has once again made a key iteration. The new Qwen3.7-Max-Preview and Qwen3.7-Plus-Preview have officially launched on the Arena AI platform and the official Qwen Chat. As the flagship preview version of the next-generation large model, these two models are expected to be officially released to the public at the upcoming Alibaba Cloud Summit.
According to the product features currently disclosed by the official sources, both new models only support the "thinking mode" in this preview phase. This means they will focus more on high-level logical reasoning and deep computing, while previously popular auxiliary tools such as web search and code interpreter are temporarily unavailable at this stage.
Excellent Performance in Official Benchmarks
In industry-recognized large model benchmark tests, the new models have shown impressive performance. In the text processing field, the comprehensive ranking of Qwen3.7 Max Preview has successfully entered the top 13 globally, which also helped Tongyi Qianwen rise to the top six in the overall laboratory ranking.
Moreover, in more specialized technical fields, the model has also demonstrated strong dominance. Data shows that it ranks 7th globally in mathematics, and it has also entered the global top ten in key sub-rankings such as expert-level applications, software and IT, and pure programming capabilities.
Breakthroughs in Visual and Expert Fields
Aside from its strong text and logical reasoning capabilities, the high-performance model Qwen3.7 Plus Preview, which was also launched this time, has also performed well in the visual field, with a comprehensive ranking of 16th globally. This achievement has elevated Tongyi Qianwen's overall laboratory ranking in the visual list to 5th globally.
