Tag
Prism ML releases Bonsai-27B-gguf, a 27-billion parameter language model with binary (1.125-bit) weights, achieving a ~14x size reduction while retaining ~90% of FP16 reasoning performance. It runs on consumer hardware with high throughput.