96 GB without HBM cost
Large-capacity GDDR7 makes local fine-tuning and large-context inference affordable for smaller teams.
NVIDIA
NVIDIA RTX PRO 6000 Blackwell 96GB GDDR7 GPU
96 GB of GDDR7 with FP4 Blackwell tensor cores — workstation and server AI without HBM-class pricing.
- Memory
- 96 GB GDDR7 ECC
- Bandwidth
- ≈1.79 TB/s
- TDP
- 600 W
- Architecture
- Blackwell (GB202)
- Precision
- FP4 / FP8 tensor
- Interface
- PCIe Gen5 x16
Need 10, 50 or 100+ units? Request volume pricing.
What Is the NVIDIA RTX PRO 6000 Blackwell?
The NVIDIA RTX PRO 6000 Blackwell is a professional GPU built on the Blackwell GB202 die with 96 GB of GDDR7 ECC memory, roughly 1.79 TB/s of bandwidth and a 600 W TDP. It supports FP4 and FP8 tensor precision over a PCIe Gen5 x16 interface, and is available in Workstation, Max-Q and passively cooled Server Edition variants.
For teams that need large memory capacity without paying HBM prices, the RTX PRO 6000 Blackwell is the pragmatic choice: 96 GB is enough to fine-tune and serve substantial models locally, and Blackwell's FP4 tensor support materially raises inference throughput for quantised models.
The three variants matter. Workstation Edition is a dual-slot blower card for desks and towers. Max-Q trades peak clocks for a lower power envelope in multi-GPU workstations. Server Edition is passively cooled for data center chassis with front-to-back airflow.
There is no NVLink on this SKU, so multi-GPU scaling goes over PCIe Gen5. That is fine for inference and many fine-tuning workloads, but large distributed training runs still favour NVLink-connected H200 or Blackwell data center parts.
NVIDIA RTX PRO 6000 Blackwell Specifications
| Manufacturer | NVIDIA |
|---|---|
| Architecture | Blackwell (GB202) |
| GPU memory | 96 GB GDDR7 with ECC |
| Memory bandwidth | ≈1.79 TB/s |
| Tensor precision | FP4, FP8, BF16, FP16 |
| TDP | 600 W (Max-Q and Server variants differ) |
| Form factor | Dual-slot blower; Max-Q and passive Server Edition available |
| Interconnect | PCIe Gen5 x16 (no NVLink on this SKU) |
| Cooling | Active blower (workstation) or passive (server) |
| Availability | Announced GTC 2025; shipping from mid-2025 |
RTX PRO 6000 Blackwell Power Consumption & Electricity Cost
Rated wall power for the NVIDIA RTX PRO 6000 Blackwell is not published in the sources we track. Contact sales for the current manufacturer specification sheet.
A verified rated wattage for the NVIDIA RTX PRO 6000 Blackwell is not published by the manufacturer in the sources we track, so we do not publish an electricity-cost estimate for it. Contact sales for the current specification sheet.
RTX PRO 6000 Cooling & Chassis Requirements
At 600 W per card, chassis selection is not a detail. Workstation Edition uses a blower that exhausts out the back of the case, which is what makes multi-card workstations viable — open-fan consumer-style coolers would recirculate heat between adjacent cards.
Server Edition ships passively cooled and depends entirely on chassis airflow. It must go into a server designed for high-static-pressure front-to-back cooling; installing it in a workstation tower will overheat it.
Plan power delivery carefully too: four cards alone is 2.4 kW before the rest of the system, which usually means redundant high-wattage PSUs and a dedicated circuit per node.
Why Consider the NVIDIA RTX PRO 6000 Blackwell?
Blackwell FP4 support
Native FP4 tensor throughput significantly accelerates quantised inference versus Hopper-generation cards.
Three cooling variants
Workstation, Max-Q and passive Server Edition cover desk-side, multi-GPU and rack deployments.
Who Is the RTX PRO 6000 Blackwell Best For?
- Local model fine-tuning and experimentation
- Quantised inference serving at FP4/FP8
- Rendering, simulation and visualisation workstations
- Teams needing large VRAM without HBM budgets
NVIDIA RTX PRO 6000 Blackwell Price
Pricing for the NVIDIA RTX PRO 6000 Blackwell can vary depending on quantity, availability, market conditions and fulfillment requirements. MinersDeals provides individual and volume quotes for qualified buyers, so we publish current pricing on request rather than as a fixed list price.
Buying multiple units? Ask about volume pricing.
Buy NVIDIA RTX PRO 6000 Blackwell
Looking to purchase the NVIDIA RTX PRO 6000 Blackwell? MinersDeals helps individual buyers, mining operators and enterprise customers source specialized hardware. Submit your requirements to receive current availability and pricing.
Wholesale & Bulk RTX PRO 6000 Blackwell Orders
Planning a larger deployment? Request pricing for 10, 50, 100 or more units. MinersDeals supports hardware procurement for mining farms, hosting providers and infrastructure operators, including staged delivery across multiple sites.
Frequently Asked Questions
How much memory does the RTX PRO 6000 Blackwell have?
96 GB of GDDR7 with ECC and roughly 1.79 TB/s of bandwidth.
Does it support NVLink?
No. This SKU connects over PCIe Gen5 x16; multi-GPU scaling happens across PCIe rather than an NVLink fabric.
What is the difference between Workstation and Server Edition?
Workstation Edition is an actively cooled blower card. Server Edition is passively cooled for data center chassis with front-to-back airflow. Core GPU and memory specifications are the same.
How does it compare with the H200?
The H200 has more memory (141 GB) and far higher bandwidth (4.8 TB/s) with NVLink, making it better for large-scale training. The RTX PRO 6000 adds FP4 support at a lower cost per card and is strong for inference and workstation use.
What is the RTX PRO 6000 Blackwell price?
Pricing varies by variant, quantity and availability. Request current enterprise pricing from MinersDeals.
Explore Related Categories
Related AI Hardware
Get RTX PRO 6000 Blackwell Pricing from MinersDeals
Tell us the quantity, destination and deployment timeline, and our team will come back with current availability and a quotation.