Hardware / NVIDIA DGX Station 748GB

NVIDIA DGX Station 748GB

748GB Unified memory / 7100 GB/s / 1600W / $85,000 MSRP / Released 2025
Find compatible models
Spec-sheet schematic of NVIDIA DGX Station 748GB - GB300 (DGX Station) chip, 748GB unified memory, 7100 GB/s
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
Chip
Nvidia GB300 (DGX Station)
Memory
748GB unified
Bandwidth
7100 GB/s
Available since
2025
Noise
Not specified

Top models (122 compatible)

GLM-5.3-Flash UD-IQ1_S Runs comfortably
93.1GB / 100.0GB min
745
TOK/S
ⓘ estimated
DeepSeek V4.1 Flash FP8 Runs comfortably
510.3GB / 515.0GB min
264
TOK/S
ⓘ estimated
DeepSeek V4 Flash 0731 UD-Q4_K_XL Runs comfortably
155.0GB / 168.0GB min
1300
TOK/S
extrapolated
Ornith-1.5-397B Q4_K_M Runs comfortably
244.0GB / 248.0GB min
16
TOK/S
ⓘ estimated
Qwen3.8-Flash-Next EXL3_2BPW Runs comfortably
62.7GB / 24.0GB min
1867
TOK/S
ⓘ estimated
See all 122 compatible models →