Qwen3.8-Max 在自主編寫程式碼與長時間執行任務上表現出色，內部測試顯示它能獨立完成為期 16 天的軟體工程專案。同時，阿里巴巴採用了僅啟動部分參數的技術來維持效率，並提供具備高度競爭力的定價策略。
分析師指出，由於美國晶片限制等因素，投資者可能低估了中國 AI 模型的實力，而阿里巴巴此次的發布證明雙方差距正在快速縮小。該消息帶動阿里巴巴股價上漲 7%，未來該公司還將開放模型權重供大眾下載自訂。
Alibaba has released its largest ever AI model, Qwen3.8-Max, claiming its performance rivals global leader Anthropic PBC. Built on 2.4 trillion parameters, the model outperforms the recently highlighted Kimi K3 on several benchmarks, demonstrating rapid progress by Chinese developers in the top-tier AI race.
Qwen3.8-Max excels in autonomous coding and long-horizon execution, having independently completed a 16-day software engineering project in internal tests. Alibaba maintains efficiency by activating only a small portion of the parameters per task and offers the model at highly competitive prices.
Analysts suggest that investors may be underestimating the capabilities of Chinese AI models due to US chip restrictions, and this release proves the gap is narrowing fast. The announcement boosted Alibaba's shares by 7%, and the company plans to release the model weights for public download soon.