Qwen3.8-Max Upgraded to Version 0902, Enhancing Programming and Office Capabilities

Alibaba Research recently released the version 0902 update for Qwen3.8-Max. Following its initial open-source release a month ago and subsequent feedback from developers and enterprises, the R&D team conducted specialized post-training focused on two key areas: code writing and collaborative office work, aiming to further enhance the model's performance in real-world business environments.

This upgrade enabled the model to achieve leading results across multiple benchmark tests, notably climbing to the top spot on the CodeArena Frontend Programming Overall Leaderboard with a total score of 1691 points. This demonstrates significant performance leaps in critical areas such as agent programming, complex multi-step logical reasoning, external tool calling, and end-to-end generation of complete applications.

While maintaining high performance, the new version significantly optimized resource consumption and pricing strategies. According to the latest model cost-performance rankings, Qwen3.8-Max-0902 incurs a comprehensive average cost of just $5 per million tokens, offering a clear cost advantage over other leading models in the same tier and helping lower the threshold for enterprise-scale deployment.

Currently, this model version is officially live on the Qwen AI Platform, providing standardized API interfaces. Meanwhile, the system has completed underlying integrations promptly, fully supporting the Qwen Office suite, Qoder development tools, and the Qwen mobile APP, delivering an instant, ready-to-use service experience for enterprise users, developers, and general consumers alike.