Unsloth publishes 1-bit GGUF quants of Mistral's 1-trillion-parameter Le Chonk

AIUnsloth released quantized GGUF versions of Mistral Large 4, a 1-trillion-parameter model, and says the 1-bit version runs locally on a Mac Studio. The post says the 1-bit build keeps 91% of top-1 accuracy across coding, agentic tasks, and math while being 80% smaller, at 223GB versus 1.1TB for FP8. The author says the quants work in Unsloth Desktop and thanks Mistral for day-zero access.

AIUnsloth says its 1-bit GGUF version of Mistral Le Chonk, a 1 trillion parameter model, runs on a Mac Studio and retains 91% top-1 accuracy. The guide lists hardware requirements from 256-270GB for 1-bit to 1.2TB for 8-bit, and the smallest 1-bit quant, UD-IQ1_S, uses about 233GB of disk space.