Alibaba's Qwen team has unveiled Qwen3.8-Max, its flagship model designed to autonomously handle complex tasks over extended periods, such as reproducing research papers and designing chips. The model scales to 2.4 trillion total parameters, with 95 billion active per query, and builds on the Qwen3.5 architecture. Alibaba first announced the model in mid-July as a preview available through Alibaba's Token Plan, Qoder, and QoderWork at ten percent of the standard price. Qwen3.8-Max is the first model in the Qwen-Max class whose weights will be made publicly available, with the team planning to release them next week. In three autonomous coding case studies, the model worked without human help: it spent 16 days building the command-line tool oh-my-cli, racking up 265 commits, 127 pull requests, and 151 issues by July 30, 2026; it reproduced the research paper "Unified Data Selection for LLM Reasoning" over roughly five days and about 125 hours of compute time, writing 7,600 lines of code and running 33 GPU training jobs, beating the paper's method on the AIME24 math benchmark by 2.7 points; and in the WWW2025 Multimodal Dialogue Intent Recognition Challenge, it fine-tuned several models within 24 hours, improving accuracy from 0.60 to 0.853 across 45