Skip to content

Qwen3.5 397B A17B on 4× H200 NVL: VRAM, speed and fit

Qwen3.5 397B A17B at 8-bit runs on the 4× H200 NVL with 256K of context at about 111 tokens a second.

Text generation · Chat · Ollama / LM Studio

Runs comfortably · on 4× H200 NVL · 8-bit (Q8_0) · Ollama / LM Studio · Chat

GPU memory
410 GB
of 564 GB
Context
256K
auto
Speed
111 tok/s
87 when full
First reply
0.3 s
2K prompt

Where the memory goes

Model
399 GB
Context
7.7 GB
Overhead
3.3 GB
OS
0.0 GB
Free
154 GB
Memory
410 GB of 564 GB
Context
256K tokens · auto
Speed
111 tok/s · 87 when full
First reply
0.3 s for a 2K prompt
Running cost
$33/mo at home · $1,162/mo on a cloud 4× MI300X

Validation: Validated

What Qwen3.5 397B A17B on 4× H200 NVL was checked against
CheckTargetResultStatus
Fits in GPU memory564 GB usable410 GB neededPass
Headroom5% free154 GB free (27%)Pass
Context for chat16K or more256KPass
Writing speed15 tok/s or more111 tok/sPass
First replyunder 3 s, 2K prompt0.3 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
Qwen3.5 397B A17B
Parameters
403B, 17B active
Precision
8-bit (Q8_0)
Context
256K, 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
$1,162/mo
4× MI300X at $9.56/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/Qwen3.5-397B-A17B-GGUF
vLLM
vllm serve Qwen/Qwen3.5-397B-A17B
MLX (Mac)
mlx_lm.chat --model mlx-community/Qwen3.5-397B-A17B-4bit

Qwen3.5 397B A17B on similar machines

All machines: Qwen3.5 397B A17B VRAM requirements. Everything that runs here: what can the 4× H200 NVL run?

Other models that run on 4× H200 NVL

Questions

Can Qwen3.5 397B A17B run on 4× H200 NVL?

Qwen3.5 397B A17B at 8-bit runs on the 4× H200 NVL with 256K of context at about 111 tokens a second.

Report VR-16BSAE9, calculated Nov 22, 2026 from published specs, not benchmarked.