Dynamic Beating AI News Flash: Alibaba's Thousand Q&A has just upgraded its flagship Qwen3.8-Max, which was released just a month ago, to introduce Qwen3.8-Max-0902. The parameter scale remains the same, with 2.4T parameters and a context of 1 million tokens. This time, the focus continues to be on post-training for Coding and Cowork, emphasizing coding, complex workflows, and long-duration agent tasks.

In Alibaba's comparison table, the 0902 version outperforms the original on all 8 coding benchmarks. TerminalBench 3.0 has increased from 11.3 to 29.0, DeepSWE 1.1 from 56.6 to 69.3, and QwenSWEbench V2 from 55.1 to 70.0. The professional task JobBench has also increased from 53.4 to 64.0.

Compared to Claude Opus 5, 0902 still loses in most coding projects, but it surpasses the opponent in MLS-Bench-Lite, SWE-Atlas QnA, and QwenSWEbench V2. In the multimodal comparison, there is only a slight increase of 0.4 to 3 points compared to the original version. It is evident that the focus of this upgrade is to make the agent more proficient in coding and task completion. The above results are currently Alibaba's internal tests.

Qwen3.8-Max-0902 is now available on the QwenCloud API, maintaining the same price of $2 for input and $6 per million tokens for output.