Unsloth AI

Headlines · 1

1. 5d
   MiniMax-M3.1-Flash Gets a 2-Bit GGUF for Local Deployment

   Unsloth AI and MiniMax report that Dynamic GGUF quantization enables local deployment of the 428B-parameter MiniMax-M3.1-Flash model. MiniMax says the 2-bit version has a 143.9GB footprint and 91.5% top-1 token agreement across seven domains, while Unsloth reports 91.5% retained top-1 accuracy, a 45% size reduction and operation on a system with 150GB of RAM. The model can be self-hosted through Unsloth Desktop and llama.cpp, with an OpenAI-compatible API available for existing applications and agents.

   UnslothAI (X) · 🔥 0

Experience and discussion from the community

Share my Unsloth AI experienceAsk about Unsloth AI