Zhipu IPO-bound, grabs code-model SOTA overnight
GLM-4.7 open-source code model beats GPT-5.1, SWE-Bench 73.8%. Try free API now.
“AI Disruption” Publication 8400 Subscriptions 20% Discount Offer Link.
The countdown to 2025 continues, with no signs of slowing in the emergence of new SOTA models.
Overnight, the coding SOTA model changed hands, and it’s open-source from launch—once again from a Chinese large model company:
Zhipu AI, GLM-4.7.
This update features coding, coding, and more coding throughout the technical report.
The most direct result of these capability improvements:
In benchmarks like AIME 2025 and Human Last Exam (HLE), GLM-4.7 scores surpass GPT-5.1;
SWE-Bench score reaches 73.8% (+5.8%), setting a new open-source record.
The official demo shows it can effortlessly create Plants vs. Zombies:
