Alibaba unveils Qwen 3.8 large language model

Alibaba announced its new large language model, Qwen 3.8, which it says contains 2.4 trillion parameters and ranks "second only to Fable 5," Anthropic’s flagship model. The company provided no model card, benchmarks, or detailed performance data, and the model’s weights are slated to be released publicly in the near future.

The launch comes amid intense competition among Chinese AI firms, with Moonshot’s Kimi K3 and GLM‑5.2 also drawing attention for strong benchmark results. Analysts note that the rapid development of open‑weight models like Qwen 3.8 could narrow the performance gap between Chinese and U.S. AI systems.