On June 13th, Zhipu AI’s GLM-5.2 opened up to all users. It’s going fully open-source under the MIT license. Developers anywhere in the world can use it, build on it, and commercialize it — for free. It supports a context window of one million tokens.

Now compare that with what happened the day before.
The U.S. Commerce Department ordered Anthropic to cut off all foreign users from two of its frontier models — Fable 5 and Mythos 5 — citing export controls. One side opening the door. The other closing it. That contrast speaks for itself.

But zoom out for a moment, because the bigger story has been building over the past month.
Chinese AI companies have been slashing API prices — in some cases by as much as 99 percent. At comparable performance levels, Chinese models now cost between one-tenth and one-thirtieth of similar Western models.
And cheaper hasn’t meant worse. China’s top models are performing at around 80 to 90 percent of the very best global systems overall. And in areas like code generation and long-context handling, some are actually ahead.
The numbers back this up. On major global developer platforms, Chinese AI models now account for over 60 percent of all API calls. Nearly half of those users are American developers — saving millions of dollars a year in compute costs alone.
Open-source. Low prices. Strong performance. Chinese AI isn’t shifting the global landscape by talking about it. It’s doing it by making the economics impossible to ignore.