Vercel CEO Guillermo Rauch was "almost shocked" by GLM-5.2's coding ability. The open-weight AI model trails Claude Opus 4.8 by just 1% on key benchmarks.

Z.ai's GLM-5.2 sits within 1% of Claude Opus 4.8 on long-horizon coding benchmarks and runs entirely on Huawei silicon.

Vercel CEO Guillermo Rauch was "almost shocked" by GLM-5.2's coding ability. The open-weight AI model trails Claude Opus 4.8 by just 1% on key benchmarks.