Urgent.News

What's breaking now, across thousands of outlets.

AI

Using AutoClaws free GLM models usage from other sdks (opencode / claude code)

ive been maintaining this small opensource project GLM Proxy (glmproxy on npm) for a while and its finally on the level where im happy to share it on reddit lmfao. it now works on windows/macos/Linux. thats been a big goal for me. honestly seeing around 5k npm downloads this past month makes me really happy. people have also helped make the project better by opening issues and sending me feedback…

I have been working on an open-source project called GLM Proxy (glmproxy on npm) for some time now, and I am thrilled to share it on Reddit. It now functions on Windows, macOS, and Linux, which was a significant goal for me. I am pleased to report that the project has garnered around 5,000 npm downloads in the past month, which fills me with great joy.

Users have also contributed to the project's improvement by opening issues and providing valuable feedback. I would like to extend my gratitude to everyone who has tried it and taken the time to reach out.

The GLM Proxy allows you to utilize the GLM models available through AutoClaw from any SDK you prefer, such as Claude Code, Cursor, Continuation, OpenCode, and other clients that support OpenAI or Anthropic APIs. To try it out, you can install it using the command "npm i -g glmproxy." You will need to have Node.js and AutoClaw installed and logged in.

Please note that it is an unofficial local proxy, so requests still go to Zhipu's cloud. There may be account and ban risks if you use it excessively in a short period, but I don't believe it will be an issue for most users. Therefore, I encourage you to read the safety notes carefully before proceeding. I have included a screenshot demonstrating its functionality across three different SDKs.

I recently decided to share this project on Reddit, as I am eager to hear about your experiences with it and to spread the word to your friends if you find it helpful. My personal use of GLM Proxy has been invaluable in my work and those of my friends, so I thought it would be worth sharing. If you appreciate the project, please consider starring it on GitHub. Your support means a great deal to me.

If you encounter any issues, please don't hesitate to let me know. I am always available to assist and make improvements. You can find the repository link here: https://github.com/eequaled/GLMproxy !

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

Local AI on four Macs: MLX capacity, speed and trust boundaries

Local AI gets a hardware question before it gets a leaderboard question: can the model and its working memory fit on the machine?

  • Four M5 Ultra Mac Studios used for Meta's Muse Spark Open model
  • Pipeline and tensor parallelism employed to manage model size
  • RDMA over Thunderbolt used for efficient tensor parallelism communication

Protecting kids from AI menace

CYBERJAYA: The misuse of artificial intelligence (AI) to create sexually explicit images, particularly of children, is on the rise, prompting calls for stronger safeguards and closer parental…

More from Sunday 18 October →