Alibaba Group has launched Qwen3.8-Max, a 2.4 trillion parameter mixture-of-experts model with approximately 95 billion active parameters. The model supports multimodal input and is available via QwenCloud, with open weights planned for release. While Alibaba reports impressive benchmark results for autonomous coding and other tasks, the article emphasizes that the practical utility of such models hinges on the robustness of their agent harnesses, including tool usage, state handling, and error recovery, rather than just parameter count or context window size. AI
IMPACT Sets a new benchmark for parameter count in open-weight models, potentially driving competition in large-scale AI development.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
