Alibaba releases Qwen3.8-Max; open weights expected next week

The 2.4T model reportedly coded autonomously for 16 days, and Alibaba claims it beats GPT-5.6 Sol while trailing only Anthropic's Fable 5. Pricing is $2/M input and $6/M output — 88% below Fable 5 on output. Open weights and a 27B variant arrive next week.
15 sources
RT @Alibaba_Qwen: Qwen3.8 is launching and going open-weight soon!🌐 With a massive 2.4T...x.com
Alibaba shares rally after unveiling its 'most powerful' AI model as U.S.-China competition heats upcnbc.com
Alibaba’s Qwen Unveils Preview of Flagship AI Modelbloomberg.com
[AINews] Qwen 3.8 Max(2.4T) and 27B, new open weights models for Coding and Coworklatent.space
Qwen 3.8 Max now available on Vercel AI Gatewayvercel.com
Qwen3.8-Max arrives with a bold claim: it outperforms GPT-5.6 Sol Max and Fable 5 on agentic computer useventurebeat.com
Alibaba outlines Qwen3.8 with 2.4 trillion parameterstechnode.com
Qwen3.8-Max: A New Bar for Coding and Coworkqwen.ai
Daniel Han of Unsloth validates Qwen3.8-27B will run only 17GB VRAMreddit.com
Qwen3.8 is launching and going open-weight soontwitter.com
Qwen by email
Get an email when Qwen has news
No news that day, no email.
More stories today
- CopilotKit open sources Channels SDK for agents in Slack and Teams
- Scoble: AI talent hot in SF, Seattle devs hit by Amazon layoffs
- Neuralink reportedly developing cognitive enhancement technology
- Post-training course materials invite educator feedback
- Rhodium's Goujon urges holistic AI safety approach