On February 9, a new pull request for integrating Qwen3.5 into the Transformers library appeared on the HuggingFace open-source project page. Industry insiders speculate that Alibaba's next-generation foundation model, Qwen3.5, is nearing its official release.

Some observers commented that this signals the start of a "crazy February" led by Chinese large language models. According to available information, Qwen3.5 utilizes a novel hybrid attention mechanism and is highly likely to be a natively visual-language model capable of visual understanding. Developers have further uncovered that Qwen3.5 may open-source at least a 2B dense model and a 35B-A3B MoE model.

Previously, The Information reported that Qwen3.5 would be open-sourced during the Chinese New Year holiday. In early February, Tang Jie, Chief Scientist at Zhipu AI, also posted on Weibo suggesting that numerous new models, including DeepSeek v4, Qwen3.5, and GLM-5, would be launched soon.