unsloth
MiniMax M2
MiniMax M2 is a 228.7B mixture-of-experts minimax-m2 model from unsloth, with about 41.5B parameters active per token. Configured context length is 196,608 tokens. License: mit. Quant sizes are the GGUF file bytes published on Hugging Face when a file exists.
Parameters229B / 41.5B active
Context192K
Attentiongqa
Licensemit
62 layers · 8 KV heads · head dim 128 · GGUF repo · config
Quant file sizes
| Quant | File size | Source |
|---|---|---|
| BF16 | 426.1 GB | Hugging Face file bytes |
| UD-Q8_K_XL | 243.4 GB | Hugging Face file bytes |
| Q8_0 | 226.4 GB | Hugging Face file bytes |
| UD-Q6_K_XL | 180.9 GB | Hugging Face file bytes |
| Q6_K | 174.9 GB | Hugging Face file bytes |
| UD-Q5_K_XL | 151.0 GB | Hugging Face file bytes |
| Q5_K_M | 151.2 GB | Hugging Face file bytes |
| Q5_K_S | 146.7 GB | Hugging Face file bytes |
| UD-Q4_K_XL | 122.6 GB | Hugging Face file bytes |
| Q4_K_M | 128.8 GB | Hugging Face file bytes |
| Q4_K_S | 121.1 GB | Hugging Face file bytes |
| IQ4_NL | 120.1 GB | Hugging Face file bytes |
| IQ4_XS | 113.5 GB | Hugging Face file bytes |
| Q4_1 | 133.4 GB | Hugging Face file bytes |
| Q4_0 | 120.6 GB | Hugging Face file bytes |
| UD-Q3_K_XL | 94.5 GB | Hugging Face file bytes |
| Q3_K_M | 101.8 GB | Hugging Face file bytes |
| Q3_K_S | 91.9 GB | Hugging Face file bytes |
| UD-IQ3_XXS | 87.2 GB | Hugging Face file bytes |
| UD-Q2_K_XL | 79.9 GB | Hugging Face file bytes |
| Q2_K | 155.3 GB | Hugging Face file bytes |
| UD-IQ2_XXS | 68.9 GB | Hugging Face file bytes |
| UD-IQ1_M | 64.0 GB | Hugging Face file bytes |
| UD-IQ1_S | 59.6 GB | Hugging Face file bytes |
On common machines
| Device | Verdict | Quant | Speed |
|---|---|---|---|
| GeForce RTX 4060 Ti 8GB | Won't fit | UD-IQ1_S | — |
| GeForce RTX 3060 12GB | Won't fit | UD-IQ1_S | — |
| GeForce RTX 4060 Ti 16GB | Won't fit | UD-IQ1_S | — |
| GeForce RTX 4090 | Won't fit | UD-IQ1_S | — |
| GeForce RTX 5090 | Won't fit | UD-IQ1_S | — |
| Apple M4 Pro | Won't fit | UD-IQ1_S | — |
| Apple M5 | Won't fit | UD-IQ1_S | — |
| CPU only · DDR5-5600 | Won't fit | UD-IQ1_S | — |
Smallest devices that fit it fully
- NVIDIA H100 SXM 80GB — Tight fit, UD-IQ2_XXS, 119–162 tok/s
- NVIDIA H100 PCIe 80GB — Tight fit, UD-IQ2_XXS, 71–97 tok/s
- NVIDIA A100 80GB SXM — Tight fit, UD-IQ2_XXS, 72–99 tok/s
- NVIDIA A100 80GB PCIe — Tight fit, UD-IQ2_XXS, 69–94 tok/s
- Apple M3 Ultra — Tight fit, UD-IQ1_M, 31–42 tok/s
- NVIDIA H200 SXM — Tight fit, UD-Q4_K_XL, 102–139 tok/s
Run it
Commands for the GeForce RTX 4060 Ti 16GB estimate (UD-IQ1_S).
llama.cpp
llama-server -hf unsloth/MiniMax-M2-GGUF:UD-IQ1_S -c 2048 -ngl 0
No matching name in the Ollama library snapshot.
LM Studio
lms get unsloth/MiniMax-M2-GGUF # Then load quant UD-IQ1_S