Skip to content

DeepSeek V3 on DGX Station GB300: VRAM, speed and fit

DeepSeek V3 at 2-bit runs on the DGX Station GB300 with 160K of context at about 150 tokens a second.

Text generation · Chat · Ollama / LM Studio

Tight fit · on DGX Station GB300 · 2-bit (Q2_K) · Ollama / LM Studio · Chat

99% used

GPU memory
251 GB
of 252 GB
Context
160K
auto
Speed
150 tok/s
111 when full
First reply
0.7 s
2K prompt

Where the memory goes

Model
239 GB
Context
10.7 GB
Overhead
937 MB
OS
0.0 GB
Free
1.3 GB
In RAM
0.0 GB
Memory
251 GB of 252 GB
Context
160K tokens · auto
Speed
150 tok/s · 111 when full
First reply
0.7 s for a 2K prompt
Running cost
$23/mo at home · $574/mo on a cloud 8× A40

Validation: Validated with conditions

What DeepSeek V3 on DGX Station GB300 was checked against
CheckTargetResultStatus
Fits in GPU memory252 GB usable251 GB neededPass
Headroom5% free1.3 GB free (1%)Check
Context for chat16K or more160KPass
Writing speed15 tok/s or more150 tok/sPass
First replyunder 3 s, 2K prompt0.7 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
DeepSeek V3
Parameters
684B, 40.1B active
Precision
2-bit (Q2_K)
Context
160K, auto
Chats at once
1
Use
Chat
Runs in
Ollama / LM Studio
Hardware
DGX Station GB300
GPU can use
252 GB of 252 GB
Spill space
480 GB LPDDR5X on the Grace CPU

Worth trying

  • 2× H200 NVL: Runs it with 160K context
  • 4× RTX PRO 5000 Blackwell 72 GB: Runs it with 160K context
  • M5 Ultra · 512 GB: Runs it with 160K context

Running cost

At home
$23/mo
1040 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 21, 2026.

How to run it

llama.cpp
llama-server -hf unsloth/DeepSeek-V3-GGUF
vLLM
vllm serve deepseek-ai/DeepSeek-V3
MLX (Mac)
mlx_lm.chat --model mlx-community/DeepSeek-V3-4bit

DeepSeek V3 on similar machines

All machines: DeepSeek V3 VRAM requirements. Everything that runs here: what can the DGX Station GB300 run?

Other models that run on DGX Station GB300

Questions

Can DeepSeek V3 run on DGX Station GB300?

DeepSeek V3 at 2-bit runs on the DGX Station GB300 with 160K of context at about 150 tokens a second.

Report VR-1UCSR3Z, calculated Nov 22, 2026 from published specs, not benchmarked.