GPU GUIDE · NVIDIA
Best AI models for the
NVIDIA DGX Station (Blackwell Ultra)
The NVIDIA DGX Station (Blackwell Ultra) has 784GB of unified memory. Below are the top 30 open-source AI models that fit, ranked by composite benchmark score. Each row shows the best quantization that fits your hardware.
VRAM
784GB
Brand
nvidia
Models that fit
30
Generation
DGX
Top 30 models for the NVIDIA DGX Station (Blackwell Ultra)
01
89.7MiMo V2.5 Pro
320BBest fit: fp16 · 641.0GBCan I run it? →
02
86.4Qwen 3.6 Max
480BBest fit: Q8_0 · 511.0GBCan I run it? →
03
85.8DeepSeek V4 Pro
685BBest fit: Q8_0 · 728.8GBCan I run it? →
04
83.3Qwen 3.6 Plus
235BBest fit: fp16 · 471.0GBCan I run it? →
05
82.9GLM-5
230BBest fit: fp16 · 461.0GBCan I run it? →
06
82.7MiniMax M2.7
456BBest fit: Q8_0 · 485.5GBCan I run it? →
07
77.5DeepSeek V4 Flash
37.0BBest fit: fp16 · 75.0GBCan I run it? →
08
76.4Qwen 3.6 27B
27.0BBest fit: fp16 · 55.0GBCan I run it? →
09
75.1Qwen 3.5 397B A17B
397BBest fit: Q8_0 · 422.8GBCan I run it? →
10
74.9MiMo V2 Omni
120BBest fit: fp16 · 241.0GBCan I run it? →
11
72.5Qwen 3.6 35B A3B
35.0BBest fit: fp16 · 71.0GBCan I run it? →
12
70.1Qwen3.5-27B
27.8BBest fit: f32 · 112.2GBCan I run it? →
13
69.5DeepSeek V3.2
685BBest fit: Q8_0 · 728.8GBCan I run it? →
14
69.3Qwen 3.5 122B A10B
122BBest fit: fp16 · 245.0GBCan I run it? →
15
65.4Mistral Medium 3.5 128B
128BBest fit: f32 · 511.8GBCan I run it? →
16
65.3Gemma 4 31B (free)
31.0BBest fit: fp16 · 63.0GBCan I run it? →
17
64.4Qwen 3.5 Omni Plus
235BBest fit: fp16 · 471.0GBCan I run it? →
18
61.9Qwen 3.5 35B A3B
35.0BBest fit: fp16 · 71.0GBCan I run it? →
19
60.0NVIDIA Nemotron 3 Super
340BBest fit: fp16 · 681.0GBCan I run it? →
20
59.9NVIDIA Nemotron 3 Super 120B A12B BF16
124BBest fit: fp16 · 248.2GBCan I run it? →
21
55.4GPT-OSS 120B
120BBest fit: fp16 · 241.0GBCan I run it? →
22
54.0Qwen 3.5 9B
9.0BBest fit: fp16 · 19.0GBCan I run it? →
23
53.8gemma 4 31B
32.7BBest fit: fp16 · 66.4GBCan I run it? →
24
52.0Gemma 4 26B A4B (free)
26.5BBest fit: f32 · 107.0GBCan I run it? →
25
49.2Qwen3 235B A22B Thinking 2507
235BBest fit: Q6_K · 194.7GBCan I run it? →
26
46.7Qwen 3 Coder
480BBest fit: Q8_0 · 511.0GBCan I run it? →
27
46.3Mistral Small 4
119BBest fit: f32 · 478.6GBCan I run it? →
28
46.1Qwen3 VL 235B A22B Instruct
236BBest fit: fp16 · 472.4GBCan I run it? →
29
45.1Qwen 3.5 4B
4.0BBest fit: fp16 · 9.0GBCan I run it? →
30
45.1DeepSeek R1 0528
685BBest fit: Q8_0 · 728.8GBCan I run it? →
Advertisement
FAQ — running AI on the NVIDIA DGX Station (Blackwell Ultra)
How many AI models can the NVIDIA DGX Station (Blackwell Ultra) run?
With 784GB of unified memory, the NVIDIA DGX Station (Blackwell Ultra) can run 30+ open-source models from our database, including MiMo V2.5 Pro, Qwen 3.6 Max, DeepSeek V4 Pro.
What's the largest LLM I can run on a NVIDIA DGX Station (Blackwell Ultra)?
The biggest model that fits is approximately 685B. Larger models would need to be quantized further or won't fit at all.
Is 784GB of unified memory enough for local AI?
Yes — 784GB comfortably runs most popular open-source models including 30B-class LLMs at Q4_K_M.