NVIDIA · Workstation · Blackwell

RTX PRO 6000 Blackwell Workstation Edition

Current2025
Specs last checked 21 Nov 2026Report an error

The most capable desktop graphics card NVIDIA sells: 188 of GB202's 192 SMs, 10% more than the RTX 5090, with three times the memory. Supports Multi-Instance GPU.

CUDA cores
24,064
Memory
96GB GDDR7
Bandwidth
1,792GB/s
AI TOPS
4,000

Overview

Brand
NVIDIA
Segment
Workstation
Series
RTX PRO Blackwell
Architecture
Blackwell
Status
Current
Launch
2025
Launch price (US)
—Not sold at a fixed MSRP. Reported at more than $8,000 per card at launch.

GPU

GPU / chip
GB202
Process
TSMC 4N
Transistors
92.2 billion
Die size
750 mm²
Streaming multiprocessors
188
CUDA cores
24,064
Tensor Cores
752
RT cores
188

Memory

Capacity
96 GB
Type
GDDR7 ECC
Bus width
512-bit
Bandwidth
1,792 GB/s

Performance

FP32 (shader)
125 TFLOPS
AI TOPS (vendor figure)
4,000

Board & platform

Power
600 W
Display outputs
4× DisplayPort 2.1b
Multi-Instance GPU
Yes
Form factor
Dual slot, double flow-through

Variants & versions

RTX PRO 6000 Blackwell Max-Q Workstation EditionSame GPU and 96 GB at 300 W with a blower cooler, for dense multi-GPU workstations.
RTX PRO 6000 Blackwell Server EditionPassively cooled data-center version used in RTX PRO Servers.

Common questions

How much memory does the RTX PRO 6000 Blackwell Workstation Edition have?

The RTX PRO 6000 Blackwell Workstation Edition has 96 GB of GDDR7 ECC on a 512-bit bus, giving 1,792 GB/s of bandwidth.

When was the RTX PRO 6000 Blackwell Workstation Edition released?

2025.