Open-source AI gains after US model restrictions

Open-source artificial intelligence models are gaining attention after US restrictions disrupted access to top closed models from Anthropic and OpenAI.

In early June, the Trump administration ordered Anthropic to block non-Americans from using its most powerful closed models, Mythos 5 and Fable 5. Faced with screening users, the company pulled the models offline entirely.

Soon afterward, OpenAI agreed to let the US government approve every customer for its newest model, GPT-5.6.

The moves have sharpened debate over closed and open AI. Closed models, including OpenAI's ChatGPT and Anthropic's Claude, are controlled by their companies and accessed through apps, websites or subscriptions. Open-source or open-weight models release core files that users can download, modify and run on their own computers.

Oren Michels, co-founder and chief executive of Barndoor AI, said relying on one frontier model can make products less reliable if access is suddenly withdrawn.

Chinese open models are among those drawing more interest. Zhipu AI, also known as Z.ai, released GLM-5.2, an open model that AI analyst Andrew Curran said is free to download, fine-tune and run on company servers.

On OpenRouter, a platform that routes requests across AI models, the combined usage share of Google, Anthropic and OpenAI fell from 55 percent to 33 percent between January and June. China's open DeepSeek now leads on the platform.

Some experts expect governments may also try to restrict open models as they become more powerful. Ethan Mollick, a University of Pennsylvania professor, said governments beyond Washington may want to keep top-tier AI locked down if models at the level of Mythos are considered risky.