- Alibaba has launched Qwen3.8-Max, calling it its 'most capable' AI model for coding and research.
- In an X post, Alibaba called Qwen 3.8-Max a new bar for coding and cowork and said its open-weight version would be released next week, allowing developers to download and modify trained parameters.
- Soon after the new models release, Alibabas Hong Kong-listed shares climbed 7 per cent.
Alibaba's Qwen 3.8-Max, launched on Monday, is touted as the company's most capable AI model, featuring 2.4 trillion parameters. This model is designed to enhance coding and research capabilities, setting a new standard in the AI landscape.
In a post on X, Alibaba stated it represents “a new bar for coding and cowork”, with an open-weight version expected next week, allowing developers to modify the model's trained parameters.23
The Qwen 3.8-Max model is built on the architecture of its predecessor, Qwen 3.5, and is said to perform on par with or better than Anthropic’s Fable 5 during tests. It supports a context window of up to 1 million tokens, enabling it to process up to 750,000 words per query.
Alibaba highlighted the model's ability to handle complex tasks, such as “designing chips nonstop for 12 hours” and “verifying protein sources”, while users engage in leisure activities. The company also noted that Qwen 3.8-Max can execute “real visual production tasks”, including creating vlogs from personal footage and developing interactive games from simple prompts.
Following the announcement, Alibaba's shares in Hong Kong surged by 7 percent, reflecting investor optimism about the new AI capabilities.
“Soon after Alibaba released the new model, its Hong Kong-listed shares climbed 7 per cent, according to CNBC. The open-weight version will allow developers to download the model's trained parameters and modify them as required.”
