Alibaba Group Holding Ltd. has previewed its most powerful artificial intelligence model, Qwen3.8, which it claims is the second most powerful model in the world, trailing only Anthropic PBC’s Claude Fable 5. The company revealed the model at the World Artificial Intelligence Conference in Shanghai.
The Qwen3.8 model accommodates 2.4 trillion parameters, making it the first Qwen model to process images, video, and documents alongside text with over 1 trillion parameters. Alibaba’s Qwen team said the model ranks “second only to Fable 5,” but it did not provide any benchmark scores to support this claim.
The absence of benchmark scores is notable, as Alibaba’s previous flagship model, Qwen3.7-Max, was released with a full set of published results, including a score of 56.6 on the Artificial Analysis Intelligence Index. In contrast, the Qwen3.8 preview arrived with no model card, no activated-parameter count, and no benchmark data.
Related: EY boosts accuracy with knowledge graphs
They have not provided a task-level comparison with Qwen3.7-Max, describing the new model only as “continuously evolving.” Alibaba has promised to release open weights “soon,” but it has not set a date or published license terms.
The Qwen3.8 preview is available now through Alibaba’s Token Plan subscription and its Qoder and QoderWork developer platforms, priced at 10% of the standard rate during the trial period. This pricing strategy may help attract developers to the new model.
Alibaba’s move puts Qwen3.8 in direct competition with a domestic rival, Chinese AI startup Beijing Moonshot AI Technology Co. Ltd., which released a 2.8 trillion-parameter model called Kimi K3 just three days earlier. Both launches push Chinese developers into the multi-trillion-parameter tier, a scale previously associated mainly with the largest U.S. labs.
Related: Cadence expands AI to circuit boards
Alibaba has made Qwen central to its effort to position itself as China’s default AI supplier. Over the past year, the company has released a steady stream of open-weight models while building out cloud infrastructure and custom chips to run them. The open-weight approach has helped Qwen assemble one of the largest developer followings of any Chinese model family.
For enterprise buyers weighing the announcement, the comparison with Fable 5 is Alibaba’s alone, and no independent leaderboard has scored Qwen3.8 yet. The last Qwen model that was ranked, Qwen3.7-Max, sits well down LMArena’s list, where Fable 5 is No. 1. Benchmarks and the promised open weights would help settle the ranking, but Alibaba has released neither.
Qwen3.8 is available for developers to test and explore.
