en.Wedoany.com Reported - Chinese AI startup Z.ai (also known as Zhupai AI) today released the GLM-5.1 open-source large model under the MIT license, allowing enterprises to download, customize, and use it commercially. The model is designed to work autonomously on a single task for up to eight hours, marking a shift from coding to agent engineering.

GLM-5.1 boasts 754 billion parameters and is designed to maintain goal alignment within extended execution trajectories spanning thousands of tool calls. Z.ai lead Lou wrote on X: "By the end of last year, agents could do about 20 steps; glm-5.1 can now do 1,700 steps. Autonomous work duration might be the most important curve after scaling laws. glm-5.1 will be the first point on this curve that the open-source community can verify firsthand." The model operates via a laddering mode to avoid plateau effects, optimizing vector databases in the VectorDBBench challenge, increasing performance from 3,547 queries per second to 21,500 queries per second, demonstrating autonomous experimentation and optimization capabilities.

On the SWE-Bench Pro benchmark, GLM-5.1 scored 58.4, surpassing GPT-5.4's 57.7 and Claude Opus 4.6's 57.3. It also demonstrated the ability to build a Linux-style desktop environment from scratch within eight hours, including a file browser, terminal, and functional games. Z.ai has integrated GLM-5.1 into the Coding Plan ecosystem, offering three subscription tiers, with API pricing set at $1.40 per million input tokens and $4.40 per million output tokens. Meanwhile, GLM-5 Turbo, a proprietary version optimized for fast inference, costs $1.20 per million input tokens and $4.00 per million output tokens.

The release of GLM-5.1 reflects AI market trends, with open-source models for broad distribution and performance-optimized variants remaining proprietary. Developer community feedback is positive, with users reporting efficiency gains, such as a task that previously took a week now being completed in just two days. Z.ai stated that the frontier of AI competition is shifting towards autonomous duration, and GLM-5.1, as an engineering-grade tool compatible with developer tools like Claude Code, is poised to transform the software development lifecycle.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com