LaunchAI ModelsAugust 3, 2026

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

Qwen3.8-Max is a 2.4-trillion-parameter MoE model, priced at $2/$6 per million tokens, with open weights promised. It claims to outperform GPT-5.6 Sol Max and Fable 5 on agentic computer use, and is now available on Vercel AI Gateway.

How this story unfolded

10 days · 5 reports · 7 community posts · from Aug 3

  1. Aug 3
  2. Aug 4
  3. Aug 6
  4. Aug 12

Daily brief

Get tomorrow's AI brief in your inbox

More stories today

Open the live feed
Alibaba Qwen releases Qwen3.8-Max, a 2.4T-parameter MoE model — AIBriefs