NVIDIA DGX Station 748GB
748GB Unified memory
/
7100 GB/s
/
1600W
/
$85,000 MSRP
/
Released 2025
Overview
NVIDIA’s desk-side datacenter machine: GB300 on a Grace Blackwell NVL architecture with 748GB of coherent unified memory at 7,100 GB/s - over 6x the M5 Ultra’s bandwidth. Shipped 2025 at $85,000. 1,600W TDP, so it needs a 20A circuit, not a wall outlet budget.
What it does well:
- Bandwidth: 7,100 GB/s decodes the largest open checkpoints faster than any consumer machine on this site - single-stream rates measured in the hundreds of tok/s on 100B-class models.
- Full-precision serving: FP8 and NVFP4 formats run natively, and 748GB holds long-context KV caches on top of full checkpoints.
Where it falls short: price and power put it out of individual reach - it is a lab and enterprise purchase. It draws more than ten times the DGX Spark’s power for a bandwidth-first trade.
Run it locally: the same DGX OS stack as the Spark, with vLLM and SGLang as the intended serving layer. The modeldex results rank checkpoints by fit and estimated tok/s.
Chip specs
- Nvidia GB300 (DGX Station)
- 748GB unified
- 7100 GB/s
- 2025
- Not specified
Top models (122 compatible)
GLM-5.3-Flash
UD-IQ1_S
Runs comfortably
93.1GB / 100.0GB min
745
ⓘ estimated
DeepSeek V4.1 Flash
FP8
Runs comfortably
510.3GB / 515.0GB min
264
ⓘ estimated
DeepSeek V4 Flash 0731
UD-Q4_K_XL
Runs comfortably
155.0GB / 168.0GB min
1300
extrapolated
Ornith-1.5-397B
Q4_K_M
Runs comfortably
244.0GB / 248.0GB min
16
ⓘ estimated
Qwen3.8-Flash-Next
EXL3_2BPW
Runs comfortably
62.7GB / 24.0GB min
1867
ⓘ estimated