Can MiniMax M2.1 run on the H100 PCIe?
Not fully in VRAM. Even at 2-bit with 8K context, MiniMax M2.1 needs about 80.3 GB, more than the 76.0 GB usable on the H100 PCIe. It can still run with offload (slower): Q4_K_M with part of the model in system RAM, about 12–23 tokens/sec (estimate). Details and command.
Runs with offload (slower)
Runs with offload 4-bit (Q4_K_M) at 8K context: all attention layers plus the experts of 34 of 62 MoE layers on the H100 PCIe (about 73.8 GB), and the experts of the other 28 layers in system RAM (about 57.0 GB). Needs a PC with at least 64 GB of system RAM. Estimated speed: 12–23 tokens/sec (usable for chat), versus a bandwidth ceiling of ~300 tok/s if it fit fully in VRAM.
llama-server -hf unsloth/MiniMax-M2.1-GGUF:Q4_K_M -c 8192 -ngl 999 --n-cpu-moe 28
Estimate, not a benchmark. The command downloads the GGUF from Hugging Face. The speed range spans dual-channel DDR4-3200 to DDR5-6000 system RAM; your CPU, RAM speed and llama.cpp version move it. If llama.cpp runs out of GPU memory, raise --n-cpu-moe by 1-2; if VRAM is left over, lower it for more speed. Recent llama.cpp builds can also pick the split automatically. How we estimate offload.
Memory by quant and context
| Quant | 4K ctx | 8K ctx | 32K ctx | 128K ctx | Max tok/s (est., 8K) |
|---|---|---|---|---|---|
| 2-bit (Q2_K) • | 79.3 GB No | 80.3 GB No | 86.1 GB No | 109 GB No | — |
| 3-bit (Q3_K_M) • | 104 GB No | 104 GB No | 110 GB No | 134 GB No | — |
| 4-bit (Q4_K_M) • | 131 GB No | 132 GB No | 137 GB No | 161 GB No | — |
| 5-bit (Q5_K_M) • | 153 GB No | 154 GB No | 160 GB No | 183 GB No | — |
| 6-bit (Q6_K) • | 177 GB No | 178 GB No | 183 GB No | 207 GB No | — |
| 8-bit (Q8_0) • | 228 GB No | 229 GB No | 235 GB No | 258 GB No | — |
| 16-bit (F16/BF16) • | 428 GB No | 429 GB No | 435 GB No | 458 GB No | — |
• real GGUF file size from unsloth/MiniMax-M2.1-GGUF. Speed is a bandwidth ceiling (2000 GB/s), not a benchmark. Methodology.
Hardware that does fit it
RTX PRO 6000 BlackwellDGX SparkRyzen AI Max+ 395 128GBMac M3 Max 128GBMac M4 Max 128GBH200
What fits instead
- Cheapest listed GPU that runs MiniMax M2.1: DGX Spark (128 GB, launch MSRP $3,999) at Q2_K, fully in VRAM at 8K context.
- Largest popular model that fits the H100 PCIe: gpt-oss 120B (117B, MoE) at F16, about 61.9 GB at 8K context.
"Runs" and "fits" mean the whole model in GPU memory at 8K context (4-bit or better for the model pick). Prices are US launch MSRPs, not today's street prices. "Popular" means the 60 most-downloaded models we track.
More
MiniMax M2.1 on every GPU and Mac · Every model that runs on the H100 PCIe · Other models that fit 64 GB VRAM · Try other settings in the calculator
Guides: Best GPU for local LLMs by budget · Can I run gpt-oss, Llama 70B or Qwen3? · Mac vs PC for local LLMs
Data: Hugging Face Hub (MiniMaxAI/MiniMax-M2.1), NVIDIA specs. Model data last checked 2026-11-17; refreshed hourly.