GLM-5.3, the AI model from Chinese startup Z.ai, scores 60 points on the Artificial Analysis Intelligence Index. That ties it with Kimi K3 for the top spot among open models, and it's seven points ahead of the previous GLM-5.2.

GLM-5.3's biggest gains come in agentic tasks. On the GDPval-AA v2 benchmark, its Elo score jumps from 1,524 to 1,770, a leap of 246 points. That puts the model in second place, behind only Claude Opus 5 (1,855). On cost, Artificial Analysis pegs GLM-5.3 at $0.68 per task, 1.5 times the $0.44 of its predecessor GLM-5.2, but 19 percent cheaper than Kimi K3 ($0.84).

The model is accessible via Z.ai’s API. However, Z.ai is delaying the release of the open weights by about two weeks because, according to the company, GLM-5.3 is so effective at detecting security vulnerabilities that it is first strengthening controls and restricting full access to select security partners.