Alibaba Qwen releases Qwen3.8-Max, a 2.4T-parameter MoE model

Qwen3.8-Max, a 2.4-trillion-parameter MoE model, accepts text, image and video input, and its open weights ship next week alongside Qwen3.8-27B. It matches Kimi K3 and DeepSeek V4 Flash on benchmarks and ranks first on Artificial Analysis' agentic index and #2 in Vision Arena.
How this story unfolded
4 days · 2 reports · 9 community posts · 11 of 12 shown
Qwen by email
Get an email when Qwen has news
No news that day, no email.
More stories today
- Reduce RAG costs on Amazon Bedrock with query-aware compression
- AWS and Panasonic Avionics use agentic AI for aircraft IFEC diagnostics
- User's Claude memory system backfired; Claude said user was the bottleneck
- Claude users discuss when to choose Sonnet over Opus
- Building token-efficient multi-agent systems