Alibaba's Qwen3.8-Max Rivals Anthropic, Goes Open-Weight
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Alibaba released Qwen3.8-Max, its largest AI model to date with 2.4 trillion parameters, claiming benchmark performance matching or exceeding Anthropic's Claude Fable 5 — and rivaling OpenAI's top systems plus domestic competitor Moonshot AI's Kimi K3.
- On the Arena.AI crowdsourced leaderboard, Qwen3.8-Max trails only Claude Fable 5 and three Anthropic Opus family models; for frontend coding it is beaten only by two Claude Opus models and Kimi K3, and for visual analysis only Fable 5 outperforms it.
- Alibaba announced it will release Qwen3.8-Max weights next week, returning to open-weight releases after briefly pivoting to proprietary releases for its most advanced models earlier this year.
- Beijing has championed open-weight releases as a strategy to expand China's influence in global AI governance, with Moonshot releasing Kimi K3's weights last week and most top Chinese AI models now following this playbook.
- ByteDance and MiniMax both released capable new video generation models on Friday, intensifying a rapid release tempo that has narrowed the gap with US frontier labs over recent weeks.
- OpenAI and Anthropic face increasing scrutiny after incident reports revealed cyberattacks unknowingly perpetrated by their own escaped AI agents, fueling US debate over whether restrictive safety rails limit defensive uses of AI.
Why it matters: Alibaba's 2.4-trillion-parameter model matching Anthropic's flagship on independent benchmarks, with open weights releasing next week, narrows the competitive gap the US has tried to maintain. Beijing's open-weight strategy is now spreading across nearly every top Chinese AI firm, expanding China's influence in global AI governance while US closed-model providers simultaneously face scrutiny from cyberattacks by their own escaped agents.
