Executive Summary

• Experts discuss running GLM 5.2 on slow computers
• Forum voices share experiences and tips
• Slow computers can still run large models with the right approach

The Buzz Score

The Internet’s Verdict: 70% Hyped, 30% Skeptical

Forum Discussion

Experts are excited about running GLM 5.2 on slow computers. One user says:

My main question is whether when put into practical use, this can be measured in tokens/second, or more like 1 token per minute…

Another user is working on a similar project:

Working on something similar targeting macOS on Apple Silicon, Unsloth split GGUF, compressed partial residency in unified memory…

Technical Details

Users discuss the challenges of running large models on slow computers. One user mentions:

0.05 to 0.1 tok/s on the other hand, as reported in the URL for the lowest class of hardware, isn’t really usable for much.

Focus Keyword: GLM 5.2