Alibaba Launches Qwen3.8-Max: 2.4T Parameters, Native Multimodal, Open Weights Next Week
mark_k · x · 2026-08-03
Alibaba has announced the release of Qwen3.8-Max, its most capable model to date. Built on the Qwen 3.5 architecture, the model features 2.4 trillion total parameters (95B active), a 1M token context window, and native multimodal support.
In a first for the series, Alibaba plans to open-source the weights of this Max-class model next week, alongside the Qwen3.8-27B model.
Key capability highlights:
- Autonomous Coding: Capable of taking real multi-day projects from an empty folder to production-ready code with minimal human intervention, utilizing self-evolving feedback loops.
- Complex Cowork: Significant gains in complex tasks involving research, long-horizon planning, and professional deliverables.
- Benchmark Jumps: Major improvements over Qwen3.7-Max on coding agents, PaperBench, OSWorld, and more.
The model is currently available via the QwenCloud API.
Related event: Alibaba releases Qwen3.8-Max, open-sources weights next week(44 posts)→
More from Models
- OpenAI stealth-testing mysterious internal model 'Aeon' in Codex desktop app and Dots — cedric_chee · 2026-10-16
- Argon edges out astra as daily driver, but Claude still wins heavy coding work — iruletheworldmo · 2026-10-16
- NVIDIA ships FP4-quantized Qwen3.8-Flash for local use, trending on Hugging Face — nvidia · 2026-10-16
- RoboVista benchmark: Gemini-4 Argon hits 81.9% on robotics understanding — berkeley_ai · 2026-10-16
- Dev reports decision model jev beats gpt-6-luna on quality in real training pipeline — genmon · 2026-10-16
- Google won't publish Argon 4 limits in AI Pro—users compare experiences — Consistent-Text-2662 · 2026-10-16