New Engine Runs 744B-Parameter AI Model on Non-Nvidia PCs

A developer created Colibri, a lightweight inference engine that runs Z.ai's GLM-5.2, a 744-billion-parameter Mixture-of-Experts model, on consumer PCs without Nvidia GPUs. The engine enables local AI inference on standard hardware, reducing dependency on specialized chips. This could democratize access to large language models for users with limited GPU resources.