Qwen3.8-Max Tops AgenticIndex Beating GPT-5.6, Weights to be Open-Sourced

千问大模型 · wechat · 2026-08-06

Alibaba's Qwen team announced that its Qwen3.8-Max model ranked first globally on ArtificialAnalysis's latest AgenticIndex with a score of 55.4, surpassing Claude Opus 5 and GPT 5.6. The model also ranked fifth on the IntelligenceIndex. The team confirmed that Qwen3.8-Max and Qwen3.8-27B will be open-sourced next week.

Related event: Alibaba's Qwen3.8-Max tops agentic benchmark, open-sources next week(11 posts)→

Original post →

More from Models

Models channel →