OPEN LINK ↗
// 4h agoMODEL RELEASE
Alibaba Cloud releases 2.4T MoE Qwen3.8-Max
Qwen3.8-Max is Alibaba Cloud's flagship 2.4-trillion parameter Sparse MoE model built for complex coding and multi-turn agentic workflows. Featuring a 1-million-token context window and native multimodal capabilities, the model is available via API and Qwen Studio for building next-generation AI agents.
// ANALYSIS
Qwen3.8-Max sets a new benchmark for frontier-class models by combining massive scale with sustained long-horizon agent execution.
- –Massive 2.4T parameter Sparse MoE design activates 95B parameters per token, balancing computational efficiency with frontier reasoning.
- –Tailored for deep software engineering with a 1M token context window capable of handling multi-file repositories and prolonged agentic sessions.
- –Native multimodal processing enables seamless translation between UI mockups, text, document streams, and video data.
// TAGS
qwenqwen3.8-maxaillmmoecodingagentic-workflowsalibaba-cloudai-codingcode-generation
DISCOVERED
4h ago
2026-08-03
PUBLISHED
10h ago
2026-08-03
RELEVANCE
9/ 10
AUTHOR
chen cheng