According to CoinMeta, Alibaba has officially released Qwen3.8-Max, which is a multimodal flagship model with 2.4T parameters, focusing on programming agent, professional office work, and long-term tasks. API has already been launched. In the four tests of terminalbench 2.1, deepswe 1.1, nl2repo, and toolathlon verified, Qwen3.8-Max scored only 1.7 to 3.9 points higher than deepseek-v4-flash. Qianwen has for the first time opened up the flagship weight level of max; the max level model weights of Qwen3.8 have been launched as hugging face and modelscope, with the model name being qwen3.8-2.4t-a95b. This model natively supports 262K contexts, can be extended to about 1M, and supports adjusting the inference intensity. The current version only supports text input, while the cloud version supports visual input and non-thinking mode. The license has been changed to a qwen custom license; ordinary users can still download, modify, and deploy it, but some large-scale commercial uses require additional authorization.