Chinese startup Z.ai's latest AI model, GLM-5.2, has made waves in Silicon Valley by ranking fourth on the Artificial Analysis Intelligence Index v4.1, outperforming leading American systems like MiniMax-M3 and DeepSeek V4 Pro. The model's coding and agentic capabilities have drawn comparisons to DeepSeek's disruptive debut in 2025.

Notably, GLM-5.2's Max tier on Code Arena's coding leaderboard secures the second position, surpassing Anthropic's Claude Opus variants. What sets Z.ai's offering apart is its pricing strategy, charging significantly lower rates than its American counterparts. At $1.40 per million input tokens and $4.40 per million output tokens, it undercuts the competition by a wide margin.

David Sacks, former White House AI czar, lauded GLM-5.2's performance, stating it is on par with models from OpenAI and Anthropic. This endorsement marks a shift from his previous stance on the US lead over Chinese labs, now acknowledging a potential rough parity in AI capabilities.

In addition to its competitive intelligence score, GLM-5.2 leads in Artificial Analysis's GDPval-AA v2 metric, showcasing its proficiency in real-world agentic tasks. Despite a trade-off in output token usage compared to peers, the model's cost advantage remains substantial. Most notably, GLM-5.2's reliance on domestic Chinese silicon, powered by Huawei processors, directly challenges US chip restrictions aimed at impeding Chinese AI advancement.