Run MiniMax-M3.1-Flash locally with Unsloth Dynamic GGUFs.
MiniMax-M3.1-Flash is MiniMax’s new 428B MoE multimodal model, with 23B active parameters and 1M context, built for coding and agentic use. Learn how to run MiniMax-M3.1-Flash via Unsloth GGUFs locally with Unsloth Desktop or llama.cpp.
Our 129.4GB 1-bit achieves ~89.9% top-1 accuracy; 143.9GB 2-bit reaches 91.5%, and 249.4 GB 4-bit UD-Q4_K_XL 98.0%. On coding, Q2 reaches 91.2% and Q4 97.6%. We weighted quant performance across seven categories: coding, math, chat, tools, multilingual, reasoning and agentic use. Thank you MiniMax for day-zero access.
See this news on the unsloth.ai official webpage.
MiniMax-M3.1-Flash is a native multimodal mixture-of-experts model for coding, agentic workflows, computer use, and cybersecurity. It supports text, image, and video inputs with a context window of up to 1M tokens.
Building on MiniMax-M3, M3.1-Flash improves task performance and inference efficiency with sparse attention throughout the model, quantization-aware training, and DSpark speculative decoding.
