How much VRAM does MiniMax M3 need?
MiniMax M3 needs 241 GB for its weights at 4-bit, 243 GB with an 8K context and 246 GB with 32K. At 8-bit, 425 GB.
- Parameters
- 427B
- Active per token
- 23B
- Longest context
- 1M
- Released
- Jun 2026
- License
- MiniMax Community
- Strongest at
- Docs
Memory at each precision
| Precision | Weights | With 8K | With 32K | With 128K |
|---|---|---|---|---|
| 16-bit | 795 GB | 797 GB | 800 GB | 811 GB |
| 8-bit | 423 GB | 425 GB | 427 GB | 439 GB |
| 6-bit | 326 GB | 328 GB | 331 GB | 342 GB |
| 5-bit | 283 GB | 285 GB | 288 GB | 299 GB |
| 4-bit | 241 GB | 243 GB | 246 GB | 257 GB |
| 3-bit | 194 GB | 196 GB | 199 GB | 210 GB |
| 2-bit | 149 GB | 151 GB | 154 GB | 165 GB |
| 1.6-bit | 82.0 GB | 84.0 GB | 86.8 GB | 98.0 GB |
One conversation, 16-bit context cache, runtime overhead included.
What runs it
| Machine | Best fit | Context | Speed | First reply |
|---|---|---|---|---|
| RTX 3060 12 GB | Too big | |||
| RTX 4060 Ti 16 GB | Too big | |||
| RTX 3090 | Too big | |||
| RTX 4090 | Too big | |||
| RTX 5090 | Too big | |||
| 2× RTX 3090 | Too big | |||
| RTX PRO 6000 Blackwell | 1.6-bit | 64K | 135 tok/s | 1.2 s |
| M4 Max 128 GB | 1.6-bit | 64K | 46 tok/s | 24 s |
| M5 Max 128 GB | 1.6-bit | 64K | 49 tok/s | 7.0 s |
| Ryzen AI Max+ 395 128 GB | 1.6-bit | 64K | 28 tok/s | 14 s |
| DGX Spark 128 GB | 1.6-bit | 256K | 35 tok/s | 4.9 s |
| M3 Ultra 256 GB | 2-bit | 256K | 41 tok/s | 14 s |
| M5 Ultra 512 GB | 6-bit | 320K | 31 tok/s | 4.3 s |
| H100 SXM | Too big | |||
| DGX Station GB300 | 3-bit | 352K | 171 tok/s | 0.4 s |
| 8× H200 NVL | 8-bit | 1M | 92 tok/s | 0.2 s |
Best fit is the highest quality, up to 8-bit, that leaves 16K of context. First reply is the wait for a 2K prompt. Speeds are rough, plus or minus 25%.
How to run it
- llama.cpp
llama-server -hf unsloth/MiniMax-M3-GGUF- vLLM
vllm serve MiniMaxAI/MiniMax-M3- MLX (Mac)
mlx_lm.chat --model mlx-community/MiniMax-M3-4bit
Questions
How much VRAM does MiniMax M3 need?
MiniMax M3 needs 241 GB for its weights at 4-bit, 243 GB with an 8K context and 246 GB with 32K. At 8-bit, 425 GB.
Can MiniMax M3 run on an RTX 4090?
Not on one card. The smallest machine here that runs it is the RTX PRO 6000 Blackwell.