Representative photo of an NVIDIA RTX professional GPU family card; illustrative only
GPU Servers
96GB GDDR7 ECC PCIe Gen5 x16 Server / Workstation Refurbished · Tested · In Stock

NVIDIA RTX PRO 6000 Blackwell 96GB

The Blackwell-generation professional card: 96GB of GDDR7 with ECC, fifth-generation Tensor Cores and fourth-generation RT Cores on a PCIe Gen5 x16 interface. Three variants - passive server edition for multi-GPU nodes, 600W workstation, and 300W Max-Q for dense four-GPU desks. Pricing on request against confirmed stock.

Request a Quote 2026 pricing: on request - quoted against confirmed stock with the date the figure holds until

Fully Tested & Documented  ·  Hong Kong Warehouse Stock  ·  Worldwide Shipping 3–7 Days  ·  1-Year Warranty

Hardware Gallery

The RTX PRO 6000, Up Close

Representative render of an NVIDIA RTX professional GPU; not the exact RTX PRO 6000 board

RTX Professional Family (Representative)

Representative photo of the NVIDIA RTX professional GPU family - the RTX PRO 6000 Blackwell is a different board. The label, serial and part number of the unit you receive are read and photographed against your order before dispatch.

Representative angle view of an RTX professional GPU card

Card Form Factor (Representative)

Representative angle view of an RTX-class professional card. The server edition is passive-cooled for multi-GPU chassis; the workstation and Max-Q editions are actively cooled for desks and dense workstation builds.

96

GB GDDR7 with ECC

600

W workstation edition (Max-Q 300W)

Gen5

PCIe x16 interface

3

variants: server, workstation, Max-Q

Key Specifications

Component Specification
ModelNVIDIA RTX PRO 6000 Blackwell - Server Edition / Workstation Edition / Max-Q Workstation Edition
ArchitectureBlackwell, fifth-generation Tensor Cores, fourth-generation RT Cores
Memory96GB GDDR7 with error-correcting code (ECC), all three variants
InterfacePCIe Gen5 x16
Power consumption400-600W (server edition); 600W (workstation); 300W (Max-Q)
CoolingPassive (server edition, for multi-GPU chassis); active double flow-through (workstation); active (Max-Q)
Form factorDual-slot: 4.4in H x 10.5in L (server and Max-Q); 5.4in H x 12in L (workstation)
Displays4x DisplayPort 2.1
Typical useEnterprise AI inference and fine-tuning, distributed rendering, HPC, virtual workstations

The RTX PRO 6000 line is one GPU in three thermal and power envelopes: the passive server edition (400-600W) is the one that goes into GPU servers four to eight cards at a time, the workstation edition runs at 600W with active cooling, and the Max-Q variant caps at 300W for dense multi-GPU workstations. Figures above are from NVIDIA's published RTX PRO 6000 Blackwell specification. For large-model serving at data-centre scale, compare with the H200; for a lower-power inference fleet, the L4. Full range: data centre GPUs.

Transparent Pricing

2026 Reference Pricing by Configuration

The three RTX PRO 6000 variants are different parts with different channel pricing, so we quote per variant on request. Send the variant, quantity and the chassis it goes into, and we confirm the fit and quote against confirmed stock.

QUOTE

Server edition

On Request

Passive, 400-600W, for multi-GPU servers. Slot, power and airflow budget checked against your chassis before quoting.

QUOTE

Workstation / Max-Q

On Request

600W active workstation or 300W Max-Q for dense four-GPU builds.

No price figure is published on this page by design. Quotations state the variant, condition and the date the figure holds until.

Origin

Hong Kong Warehouse

Free port — full export documentation

Transit

3–7 Days Air Freight

Tracked, insured, worldwide

Warranty

1-Year Warranty

Diagnostic report in every box

Quality Process

What We Test Before an RTX PRO 6000 Ships

1

Memory and ECC

All 96GB exercised with ECC telemetry logged, including a sustained load long enough to raise junction temperatures and expose marginal stacks.

2

Variant and Power Profile

Server, workstation and Max-Q variants are distinguished by power limit, cooler and part number - the variant you ordered is confirmed on the label, not assumed from the photo.

3

PCIe Gen5 Link

Link trained at Gen5 x16 on a reference platform, with width and rate logged; cards that only train at Gen4 or x8 are rejected.

4

Thermal and Fan Behaviour

For actively cooled variants, fan curves and thermal throttling points are logged under sustained load; for passive server cards, the board is tested in a server airflow path.

5

Provenance

Serial and part number verified against board markings; mining-modified boards and reflashed BIOS are screened out and never sold as professional parts.

Report in the Box

Every unit ships with its per-machine diagnostic report and warranty. Full process: inspection guide.

Frequently Asked Questions

What is the difference between RTX PRO 6000 Server and Workstation editions?

Cooling and power envelope. The server edition is passively cooled for multi-GPU chassis and runs at 400-600W; the workstation edition is actively cooled, double flow-through, at 600W; the Max-Q edition caps at 300W for dense four-GPU workstations. Memory (96GB GDDR7 ECC) and the PCIe Gen5 x16 interface are the same across the three.

Is the RTX PRO 6000 a data centre GPU or a workstation card?

Both, depending on the variant. The server edition is qualified for multi-GPU server deployment and is the one we normally quote for inference nodes; the workstation and Max-Q editions are for desks and dense workstation builds. It is not a replacement for H200-class NVLink training systems - it has no NVLink fabric and a different memory hierarchy.

How much does an NVIDIA RTX PRO 6000 cost?

Pricing is on request against confirmed stock, with the date the figure holds until stated in writing. Variant matters: server, workstation and Max-Q editions are different parts with different channel pricing, so tell us which one the build requires.

Can I put four RTX PRO 6000 cards in one server?

Yes, with the right chassis: the passive server edition is designed for that, but you need the power budget (up to 600W per card), adequate airflow and PCIe bifurcation/width to feed each card at Gen5 x16. Send the server model and we check the slot, power and cooling budget before quoting.

What condition and warranty do you offer?

New and tested pre-owned units, each with a per-serial test record (memory/ECC, sustained load, PCIe link training, thermal behaviour) and a 1-year written warranty. Condition is stated line by line before you commit.

Request an RTX PRO 6000 Quote

Tell us which variant (server, workstation, Max-Q), quantity and destination. We confirm stock and the fit for your chassis, then quote against confirmed supply with the date the figure holds until.

Request a Quote