Bonsai 27B by Prism ML is a 27 billion parameter dense language model quantized to 1-bit precision (3.8 GB vs. original 54 GB). Inference runs locally on your GPU through WebGPU. Your prompts and replies stay on your device.
Enter a Hugging Face access token with read access to prism-ml/Bonsai-27B-gguf. It stays in this browser — requests go only to huggingface.co.
The entire 27 billion parameter model runs entirely on your device. No data ever leaves your machine.
Kernels are the low-level GPU programs that do the model's actual math — the 1-bit matrix multiplies, linear attention and normalizations behind every token. Each one is a WGSL compute shader, compiled by your browser for this GPU.