No machine on this network has measured this model yet. Everything below is what the catalog knows exists — where the weights come from, how large each build is, and whose work it is. Nothing here is a claim about what the model can do; that only appears once a machine has run it and signed the result.
unsloth/MiniMax-M3-GGUF:IQ1_Mllamacpp · IQ1_M · 120.4 GiB · quantizer (unsloth)
unsloth/MiniMax-M3-GGUF:IQ2_XXSllamacpp · IQ2_XXS · 125.6 GiB · quantizer (unsloth)
unsloth/MiniMax-M3-GGUF:IQ3_XXSllamacpp · IQ3_XXS · 149.3 GiB · quantizer (unsloth)
pipenetwork/MiniMax-M3-MLX-3bitmlx · 3BIT · 173.6 GiB · quantizer (pipenetwork)
spectator2026/MiniMax-M3-AWQ-int4vllm · W4A16 · 224.3 GiB · quantizer (spectator2026)
mlx-community/MiniMax-M3-4bitmlx · 4BIT · 224.9 GiB · quantizer (mlx-community)
unsloth/MiniMax-M3-GGUF:Q4_K_Sllamacpp · Q4_K_S · 231.5 GiB · quantizer (unsloth)
nvidia/MiniMax-M3-NVFP4vllm · NVFP4 · 232.9 GiB · quantizer (nvidia)
unsloth/MiniMax-M3-GGUF:Q4_K_Mllamacpp · Q4_K_M · 246.7 GiB · quantizer (unsloth)
EmbeddedLLM/MiniMax-M3-FP8-dynamicvllm · FP8 · 401.1 GiB · quantizer (EmbeddedLLM)
MiniMaxAI/MiniMax-M3-MXFP8vllm · MXFP8 · 413.3 GiB · vendor (MiniMaxAI)
pipenetwork/MiniMax-M3-MLX-8bitmlx · 8BIT · 421.5 GiB · quantizer (pipenetwork)
unsloth/MiniMax-M3-GGUF:Q8_0llamacpp · Q8_0 · 422.5 GiB · quantizer (unsloth)
unsloth/MiniMax-M3-GGUF:BF16llamacpp · BF16 · 794.3 GiB · quantizer (unsloth)
MiniMaxAI/MiniMax-M3vllm · BF16 · 795.5 GiB · vendor (MiniMaxAI)