Open-weight models hit a tipping point. GLM 5.2 runs as a real subagent in production. Not as a demo. Not as a benchmark screenshot. But as a near-frontier perf

Open-weight models hit a tipping point. GLM 5.2 runs as a real subagent in production. Not as a demo. Not as a benchmark screenshot. But as a near-frontier performance you can actually deploy. [📷](https://pbs.twimg.com/media/HMyTbUwaoAA--o5.jpg)