At a Glance · AI-generated

Alibaba released Qwen3.8-Max, a 2.4 trillion parameter mixture-of-experts model on August 3, marking the most capable offering in its Qwen family. By August 6, benchmarking comparisons showed the model performing competitively with Claude Opus 4.8 while trailing Kimi K3, which achieves higher scores at a 25 percent cost reduction. The release underscores intensifying competition in large language model development among Chinese and global AI developers.