GLM-5.1 Launch: Coding Power Near Claude Opus 4.6, Subscriptions Sell Out

GLM-5.1, the latest model from Zhipu, is now publicly available for all GLM Coding Plan users and is positioned as a coding‑focused large language model.

In the Coding Evaluation benchmark the model achieves 45.3 points, a rise of nearly 10 points from GLM‑5’s 35.4, and trails the leading Claude Opus 4.6 by only 2.6 points, leading observers to speculate that GLM‑5.1 may soon surpass Sonnet 4.6.

The model retains a 200 K token context window, inherits GLM‑5’s configuration, and adds a reasoning mode. It can be switched to Claude Code, integrated with OpenClaw custom settings, and is compatible with the OpenAI API, lowering the integration barrier for developers.

User tests demonstrate that GLM‑5.1 can generate well‑structured technical manuals, create interior‑design floor plans, build an interactive chess web game, and even produce a Kandinsky‑style Minecraft‑like clone, showing strong spatial‑structure understanding, consistency, and dynamic completion.

GLM‑5.1 was released just over a month after GLM‑5, making it a rapid‑iteration, reinforced version that improves stability, execution efficiency, and handling of complex tasks.

Following the launch, all three GLM Coding Plan tiers—Lite, Pro, and Max—sold out immediately. The Max tier will be restocked on March 28 at 10 am, with each tier offering different quotas and benefits to match various development needs.

For macOS users, the official guide describes how to configure GLM‑5.1 in Claude Code by editing the configuration file and replacing environment variables; the /status command can verify the model’s state. Currently only the Coding Agent that supports custom models can use GLM‑5.1, while other tools await future support.

Further details and documentation are available at the official URLs.