
Alibaba Unveils Qwen3.8-Max, Its Most Capable AI Model Yet
The story in brief
Alibaba unveiled Qwen3.8-Max, its largest AI model, outperforming Fable 5 on PaperBench (93.0 vs. 88.8) and ERQA (77.8 vs. 70.0). It scored 52.4 on Agents' Last Exam, slightly behind GPT-5.6 Sol's 53.6. On TerminalBench-2.1, it scored 86.6, ahead of Fable 5's 84.6. It performed well on BabyVision and CharXiv but lagged Fable 5 in Python-assisted visual tasks. Alibaba's AI strategy includes a mix of approaches.