China turns up the heat with open model blitz as US model makers panic
For the first time, Alibaba's Qwen team is letting its 'Max' model out of the API pen; meanwhile, DeepSeek V4-Flash gives new meaning to cheap and cheerful competition
The AI arms race intensified on Monday as Chinese e-commerce and cloud provider Alibaba unveiled Qwen 3.8-Max, a 2.4 trillion-parameter model that competes with top models from Anthropic and OpenAI. This follows the launch of DeepSeek V4 Flash 0731, which benchmarked just slightly behind OpenAI's budget-friendly GPT-5.6 Luna while costing 40% less per task.
Chinese model developers are now targeting American counterparts in both price and performance. US model developers, including Anthropic and OpenAI, are raising concerns over the origins and safety of China-made AI models, despite their own safety concerns. Alibaba, a major player in the Chinese market, is now releasing its most capable model weights for download, Qwen 3.8-Max.
The model requires substantial GPU resources for deployment, with internal workloads needing 8-16 GPUs. However, Alibaba plans to release a lighter 27-billion parameter version. Analysts suggest that China is currently leading the open weights arena, and could potentially dominate the frontier models by the end of this year or next year.
Written by urgent.news from The Register Science's reporting — not their text. Machine-written — it may contain errors, so check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
