Skip to content

Qwen3.5 397B A17B on 4× RTX 3090: VRAM, speed and fit

Qwen3.5 397B A17B at 1.6-bit runs on the 4× RTX 3090 with 256K of context at about 108 tokens a second.

Text generation · Chat · Ollama / LM Studio

Runs comfortably · on 4× RTX 3090 · 1.6-bit (IQ1_S) · Ollama / LM Studio · Chat

GPU memory
88.5 GB
of 96.0 GB
Context
256K
auto
Speed
108 tok/s
46 when full
First reply
1.0 s
2K prompt

Where the memory goes

Model
77.5 GB
Context
7.7 GB
Overhead
3.3 GB
OS
0.0 GB
Free
7.5 GB
Memory
88.5 GB of 96.0 GB
Context
256K tokens · auto
Speed
108 tok/s · 46 when full
First reply
1.0 s for a 2K prompt
Running cost
$20/mo at home · $131/mo on a cloud 4× RTX A5000

Validation: Validated

What Qwen3.5 397B A17B on 4× RTX 3090 was checked against
CheckTargetResultStatus
Fits in GPU memory96.0 GB usable88.5 GB neededPass
Headroom5% free7.5 GB free (8%)Pass
Context for chat16K or more256KPass
Writing speed15 tok/s or more108 tok/sPass
First replyunder 3 s, 2K prompt1.0 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
1.6-bit (IQ1_S)
Context
256K, auto
Chats at once
1
Use
Chat
Runs in
Ollama / LM Studio
Hardware
4× RTX 3090
GPU can use
96.0 GB of 96.0 GB

Running cost

At home
$20/mo
920 W at the wall
Renting instead
$131/mo
4× RTX A5000 at $1.08/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

  • RTX PRO 5000 Blackwell 48 GBToo big
  • RTX 6000 AdaToo big
  • L40SToo big
  • 2× RTX 5090Just misses
  • RTX PRO 6000 BlackwellRuns
  • Mac mini M4 16 GBToo big
  • Mac mini M4 24 GBToo big
  • Mac mini M4 32 GBToo big

All machines: Qwen3.5 397B A17B VRAM requirements. Everything that runs here: what can the 4× RTX 3090 run?

Other models that run on 4× RTX 3090

Questions

Can Qwen3.5 397B A17B run on 4× RTX 3090?

Qwen3.5 397B A17B at 1.6-bit runs on the 4× RTX 3090 with 256K of context at about 108 tokens a second.

Report VR-19YYTG2, calculated Nov 22, 2026 from published specs, not benchmarked.