Dell Pro Max Desktop (NVIDIA GB300)

Linux

Dell Pro Max Desktop with NVIDIA Grace Blackwell Ultra GB300 Superchip (288GB HBM3e), 496GB LPDDR5X system RAM, 72-core Grace CPU. Includes RTX Pro 2000 Blackwell (16GB GDDR7) for display. Liquid-cooled, 1600W Titanium PSU. Runs Ubuntu / NVIDIA DGX OS — not a consumer Windows PC.

From

$30,000

Estimated · varies by configuration

Enterprise pricing varies by configuration and region. Confirm quote and availability with Dell.

View on Dell

Memory

288 GB

GPUs

1×

RAM

496 GB

Models

Type

Desktop

Inference Memory

288 GB

Accelerator

288 GB HBM3e

System RAM

496 GB

CPU

NVIDIA Grace 72-core Arm Neoverse V2

Linux

What it can run

64 models


all-MiniLM-L6-v2	FP16	3000 tok/s	Excellent
Arcee Trinity Large Thinking 400B	Q4_K_M	41 tok/s	Excellent
Arcee Trinity Mini 26B	Q8_0	332 tok/s	Excellent
Arcee Trinity Nano 6B	Q8_0	1411 tok/s	Excellent
Code Llama 34B Instruct	Q5_K_M	135 tok/s	Excellent
Codestral 22B	Q5_K_M	180 tok/s	Excellent
Command R 35B	Q8_0	110 tok/s	Excellent
DeepSeek Coder V2 Lite 16B	Q8_0	210 tok/s	Excellent
DeepSeek R1	Q4_K_M	20 tok/s	Good
DeepSeek R1 Distill Qwen 32B	Q8_0	120 tok/s	Excellent
DeepSeek R1 Distill Qwen 7B	Q8_0	360 tok/s	Excellent
DeepSeek V3	Q4_K_M	22 tok/s	Good
FLUX.1 Dev	FP16	15 tok/s	Excellent
Gemma 2 27B Instruct	Q5_K_M	145 tok/s	Excellent
Gemma 2 9B Instruct	Q8_0	320 tok/s	Excellent
Gemma 3 12B	Q8_0	250 tok/s	Excellent
Gemma 3 27B	Q8_0	130 tok/s	Excellent
Gemma 3 4B	Q8_0	500 tok/s	Excellent
Gemma 4 26B-A4B	Q8_0	500 tok/s	Excellent
Gemma 4 31B	Q8_0	183 tok/s	Excellent
Gemma 4 E2B	Q8_0	500 tok/s	Excellent
Gemma 4 E4B	Q8_0	500 tok/s	Excellent
GigaChat Lightning 10B	Q8_0	320 tok/s	Excellent
InternLM 2.5 7B Chat	Q8_0	350 tok/s	Excellent
Llama 3.1 70B Instruct	Q5_K_M	65 tok/s	Excellent
Llama 3.1 8B Instruct	Q8_0	350 tok/s	Excellent
Llama 3.2 11B Vision	Q8_0	260 tok/s	Excellent
Llama 3.2 1B Instruct	Q8_0	800 tok/s	Excellent
Llama 3.2 3B Instruct	Q8_0	650 tok/s	Excellent
Llama 3.3 70B Instruct	Q8_0	55 tok/s	Excellent
Llama 4 Scout	Q8_0	40 tok/s	Excellent
LLaVA 1.6 13B	Q5_K_M	270 tok/s	Excellent
Mistral 7B Instruct v0.3	Q8_0	380 tok/s	Excellent
Mistral Large 2 123B	Q8_0	30 tok/s	Good
Mistral Small 24B Instruct	Q8_0	150 tok/s	Excellent
Mixtral 8x7B Instruct	Q5_K_M	100 tok/s	Excellent
nomic-embed-text v1.5	FP16	2000 tok/s	Excellent
NVIDIA Nemotron-3-super-120B-A12B	Q4_K_M	180 tok/s	Excellent
Phi-3 Medium 14B Instruct	Q8_0	230 tok/s	Excellent
Phi-3 Mini 3.8B Instruct	Q8_0	550 tok/s	Excellent
Phi-4 14B	Q8_0	220 tok/s	Excellent
Phi-4 Mini	Q8_0	580 tok/s	Excellent
Qwen 2.5 14B Instruct	Q8_0	220 tok/s	Excellent
Qwen 2.5 72B Instruct	Q4_K_M	60 tok/s	Excellent
Qwen 2.5 7B Instruct	Q8_0	360 tok/s	Excellent
Qwen 2.5 Coder 32B Instruct	Q5_K_M	140 tok/s	Excellent
Qwen 2.5 Coder 7B Instruct	Q8_0	360 tok/s	Excellent
Qwen3-14B Instruct	Q8_0	230 tok/s	Excellent
Qwen3-30B-A3B	Q8_0	145 tok/s	Excellent
Qwen3-32B Instruct	Q8_0	120 tok/s	Excellent
Qwen3-8B Instruct	Q8_0	340 tok/s	Excellent
Qwen3.5-122B-A10B	Q8_0	200 tok/s	Excellent
Qwen3.5-27B	Q8_0	150 tok/s	Excellent
Qwen3.5-397B (MoE)	Q4_K_M	120 tok/s	Excellent
Qwen3.6-27B	Q8_0	150 tok/s	Excellent
Qwen3.6-35B-A3B	Q5_K_M	145 tok/s	Excellent
QwQ 32B Preview	Q5_K_M	140 tok/s	Excellent
Stable Diffusion 3 Medium	FP16	20 tok/s	Excellent
Stable Diffusion 3.5 Large	FP16	12 tok/s	Excellent
Stable Diffusion XL 1.0	FP16	18 tok/s	Excellent
StarCoder 2 15B	Q8_0	210 tok/s	Excellent
Whisper Large V3	FP16	450 tok/s	Excellent
Whisper Large V3 Turbo	FP16	600 tok/s	Excellent
Yi 1.5 34B Chat	Q8_0	110 tok/s	Excellent

Showing 64 of 64 entries

Best Fit

Who this machine makes sense for

This machine is aimed at team, lab, or enterprise buyers who want a supported system instead of assembling a tower. 288 GB makes it viable for serious local workloads without a DIY build process.

Before You Buy

What to verify first

The main question is not whether the machine works, but whether the price premium is justified by warranty, support, and deployment simplicity versus an equivalent custom build.

All machines