China's AI Prowess Grows: Qwen3.8-Max Challenges US Dominance

Alibaba has introduced Qwen3.8-Max, its most advanced AI model to date, equipped with 2.4 trillion parameters. This powerful model specializes in long-horizon agentic tasks, operating autonomously for extended periods, demonstrated by a recent 125-hour experiment. Internally, the model successfully replicated a research paper's experiment, generating all necessary code and analysis from scratch.

Test results show Qwen3.8-Max performing comparably to leading models from OpenAI and Anthropic, even surpassing them in visual reasoning and agentic computer use benchmarks. Alibaba plans to release the model with open weights next week. This release follows recent advancements from other Chinese AI labs, increasing concerns in the U.S. about a potential Chinese lead in AI development.

The cost-effectiveness of Qwen3.8-Max is also a significant factor, priced at $2 per million input tokens and $6 per million output tokens, a fraction of the cost of competitors like Fable 5. Public perception of AI differs between China and the U.S., with a greater acceptance and optimism towards AI in China compared to prevalent anxiety in the United States. Promotional materials from both sides reflect these contrasting sentiments, with Alibaba's ad promoting a relaxed, AI-assisted lifestyle and U.S. companies addressing public fears.