Skip to content

MiniMax M2.7 on 4× H200 NVL: VRAM, speed and fit

MiniMax M2.7 at FP8 runs on the 4× H200 NVL with 200K of context at about 149 tokens a second.

Text generation · Chat · Ollama / LM Studio

Runs comfortably · on 4× H200 NVL · FP8 · Ollama / LM Studio · Chat

GPU memory
266 GB
of 564 GB
Context
200K
auto
Speed
149 tok/s
45 when full
First reply
0.2 s
2K prompt

Where the memory goes

Model
214 GB
Context
48.4 GB
Overhead
3.0 GB
OS
0.0 GB
Free
298 GB
Memory
266 GB of 564 GB
Context
200K tokens · auto
Speed
149 tok/s · 45 when full
First reply
0.2 s for a 2K prompt
Running cost
$33/mo at home · $574/mo on a cloud 8× A40

Validation: Validated

What MiniMax M2.7 on 4× H200 NVL was checked against
CheckTargetResultStatus
Fits in GPU memory564 GB usable266 GB neededPass
Headroom5% free298 GB free (53%)Pass
Context for chat16K or more200KPass
Writing speed15 tok/s or more149 tok/sPass
First replyunder 3 s, 2K prompt0.2 sPass

Targets: a reply you can read along with, a first reply the use case tolerates, and 5% of memory left free. Calculated from published specs, not benchmarked.

The setup

Model
MiniMax M2.7
Parameters
229B, 10B active
Precision
FP8
Context
200K, auto
Chats at once
1
Use
Chat
Runs in
Ollama / LM Studio
Hardware
4× H200 NVL
GPU can use
564 GB of 564 GB

Running cost

At home
$33/mo
1520 W at the wall
Renting instead
$574/mo
8× A40 at $4.72/h

4 h a day, power at $0.18 per kWh, cloud at RunPod on-demand prices as of Nov 22, 2026.

How to run it

llama.cpp
llama-server -hf unsloth/MiniMax-M2.7-GGUF
vLLM
vllm serve MiniMaxAI/MiniMax-M2.7
MLX (Mac)
mlx_lm.chat --model mlx-community/MiniMax-M2.7-4bit

MiniMax M2.7 on similar machines

All machines: MiniMax M2.7 VRAM requirements. Everything that runs here: what can the 4× H200 NVL run?

Other models that run on 4× H200 NVL

Questions

Can MiniMax M2.7 run on 4× H200 NVL?

MiniMax M2.7 at FP8 runs on the 4× H200 NVL with 200K of context at about 149 tokens a second.

Report VR-0YN3SYI, calculated Nov 22, 2026 from published specs, not benchmarked.