Alibaba touts Qwen3.8 as No. 2—with no proof yet

Alibaba has unveiled Qwen3.8, a 2.4-trillion-parameter model it calls its most powerful yet — and says it is “second only to Fable 5.” The claim, made by the company’s Qwen team on X during the World Artificial Intelligence Conference in Shanghai, arrived without the usual evidence. According to SiliconANGLE, Alibaba published no benchmark scores, no model card, and no independent test results to support the ranking.

Alibaba shares rose as much as 5.4% on Monday after the announcement.

Qwen3.8 is the first model in the Qwen family above 1 trillion parameters to be multimodal, meaning it can work with images, video, documents, and text. Alibaba says it outperforms its previous flagship, particularly in coding and office tasks such as full-stack development and data analysis. Lead developer Shuai Bai described it as the team’s first trillion-parameter multimodal model, with understanding he said matches or exceeds leading proprietary systems.

Recommended reading

Jaron Lanier says 'AI' is the wrong story

The preview is available now through Alibaba’s Token Plan subscription and Qoder developer tools, priced at 10% of the standard rate during the trial. Alibaba says full open weights are coming “soon,” but it has not given a date or disclosed licence terms.

Missing benchmarks and open-weight timing

That lack of evidence marks a change for Alibaba. Its previous flagship, Qwen3.7-Max, launched in May with a full set of published results. Qwen3.8 has launched with none, and no public leaderboard has ranked it yet. On LMArena, where Fable 5 holds the top spot, the older Qwen3.7-Max sits much lower.

Outside verification will only be possible once the weights are released. Until then, Alibaba’s comparison remains its own.

China’s latest open-model push

The timing appears deliberate. The launch came three days after Moonshot released Kimi K3, a 2.8-trillion-parameter open model that topped a major coding leaderboard and shook US chip stocks, as The Decoder noted. Other Chinese labs are also pushing at the frontier, including Zhipu with GLM-5.2.

There is also a business angle: like Moonshot, Alibaba is putting the model into paid products first and releasing the weights later, capturing customer demand before free access arrives. That matters because Alibaba’s biggest models have not typically been open-weight. Nathan Lambert said the move signals a new phase in competition among Chinese labs.

If the eventual release matches Alibaba’s claims, it would bring open-weight models deeper into territory usually dominated by the biggest US labs. For now, though, the company is asking the market to trust the hype ahead of the download.