Zhipu AI launches open-source GLM-5.1 model for coding tasks
SMRTR summary
Eight hours. That's how long Z.ai's new GLM-5.1 model can reportedly stay focused on a single coding project, methodically planning, executing, testing, and refining code before delivering production-ready results.
This Chinese AI company has unveiled what it calls the first model from China capable of such marathon coding sessions, marking a significant shift from traditional one-shot code generation to sustained autonomous work.
The technical benchmarks tell an impressive story. GLM-5.1 scores 58.4 on SWE-Bench Pro, outperforming established rivals like GPT-5.4 and Claude Opus. But Z.ai isn't just chasing numbers on paper.
The model is already live for developers using GLM Coding Plans, with integration guides for popular tools like Claude Code and Cline. The company has also released downloadable model weights, signaling confidence in their breakthrough.
Z.ai, formerly known as Zhipu AI, represents one of China's rising AI powers that completed a Hong Kong listing this year. Their focus on extended autonomous coding suggests the next frontier in AI development may not be flashier chatbots, but patient, persistent digital engineers capable of tackling complex, multi-hour programming challenges.
SMRTR provides this summary for quick context. The original article belongs to Testing Catalog.
Read the original article