China's Open-Weight AI Models Close the Gap with American Frontier Labs

Moonshot and Alibaba have released open-weight models that rival OpenAI and Anthropic at a fraction of the cost, accelerating a shift that could reshape the global AI landscape.

Moonshot's Kimi K3 and Alibaba's Qwen 3.8 have reached performance levels competitive with frontier models from OpenAI and Anthropic, according to benchmarks and early user reports. The releases came within days of each other and represent a significant milestone: open-weight models from China are now effectively matching American proprietary systems.

The timing underscores a strategic divergence. While American labs keep their best models behind APIs and subscription paywalls, Chinese companies are releasing weights openly — a move that turns the US export controls on GPUs into a distribution advantage. "There's an 80% chance that any given startup is using Chinese models," a16z partner Martin Casado told The Economist, citing the permissionless nature of open-weight releases.

Ben Werdmuller, writing at werd.io, argues that AI models as products have very little moat beyond brand loyalty. "It's easy to switch between them: someone could be using ChatGPT today and Claude tomorrow, with very little impact on their workflows." When the product layer is commoditized, open distribution wins — and China is betting big on that dynamic.

The economic stakes are substantial. AI infrastructure spending is a major driver of the US economy right now. If the bottom falls out of that spending as open-weight models undercut proprietary pricing, the ripple effects could be severe. The gap is closing faster than many expected — and the open-weight playbook is proving harder to compete with than export controls.

Sources: The Verge werd.io