China's Moonshot and Alibaba unveil giant AI models they say rival GPT-5 and Claude
Two of China's biggest AI players released models this weekend claiming to match the best American systems, with plans to make them free to download. The gap between US and Chinese AI may be closing faster than Washington expected.
Key points
- Moonshot AI, a Beijing-based lab, released Kimi K3 on Friday, July 18, 2025, claiming it ranks above nearly every US model except OpenAI's GPT-5.6 Sol and Anthropic's Claude Fable 5.
- Alibaba previewed Qwen3.8 over the weekend, calling it "second only to Fable 5" and one of the most powerful models available today.
- Moonshot says Kimi K3 has 2.8 trillion parameters, the internal numbers a model learns during training, making it the largest open-source AI system by that measure.
- Both companies plan to release their model weights, the files developers need to run and modify an AI model themselves, to the public for free.
- US export controls have blocked China from buying the most advanced chips, yet these models appear to approach the American frontier anyway.
Two of China's top AI companies dropped major announcements this weekend, and the timing wasn't accidental.
Moonshot AI, based in Beijing, unveiled Kimi K3 on Friday. Alibaba followed days later with a preview of Qwen3.8. Both say their models sit just behind the very best American systems, a claim that, if it holds up under independent testing, shows how fast the gap between US and Chinese AI is closing. We first reported Kimi K3's ambitions on 16 July in our earlier story on what it might mean for Anthropic's lead.
The comparisons both companies chose are pointed. Moonshot says Kimi K3 trails only OpenAI's GPT-5.6 Sol and Anthropic's Claude Fable 5 in its own tests. Alibaba calls Qwen3.8 "second only to Fable 5." Neither OpenAI nor Anthropic publish technical specifications for their leading models, so independent verification is hard.
Size is part of the story. Moonshot puts Kimi K3 at 2.8 trillion parameters. Alibaba puts Qwen3.8 at 2.4 trillion. Parameters are a rough guide to scale, not a guaranteed measure of quality: a smaller, well-trained model can beat a larger one on many tasks.
Why should ordinary people care about this?
If these models are released as promised, developers anywhere in the world can download them, run them on their own machines, and build products on top of them for free. That changes who gets to build AI tools and at what cost.
Moonshot says it will publish the full model weights on July 27, 2026. Alibaba says Qwen3.8 is "going open-weight soon." Once those files are public, no company controls them.
This openness contrasts sharply with leading US labs, which keep their most powerful models proprietary. Washington has also used export controls to restrict China's access to the advanced chips, called GPUs, that power AI training. Those restrictions haven't stopped these releases.
First reported by The Verge AI, the announcements echo the shock of DeepSeek last year, when a Chinese lab released a capable model built at surprisingly low cost. The assumption that enormous US spending on chips and data centres guaranteed a lasting edge took a serious hit then. It's under pressure again now.
For anyone building AI-powered products, the practical consequence is real: two more powerful base models are about to be free. That expands your options and drives down the cost of entry, wherever you are.
