Instead, this Pro-labelled GPU line is aimed squarely at graphics-heavy professional workloads, not just AI.

The RTX Pro 6000 has three flavours: a Server Edition with passive cooling, a chunky Workstation Edition with a desktop-style cooler, and a Max-Q variant using a blower-style setup ideal for packing multiple cards into one system.

All three come with 96GB of ECC GDDR7 memory, beating the pants off Nvidia’s H100 NVL in raw capacity. They also sport RTX cores, making them viable for high-end rendering, simulation, and not just endless AI training sessions. The Server version ranges from 400 to 600W, while the workstation model gobbles a fixed 600W. The Max-Q design is built for clusters of four GPUs in one rig.

This family is turning heads not just for specs, but for pricing. Unlike data centre monsters like the H100, H200, or the newer B200, the RTX Pro 6000 is more accessible. It doesn’t rely on HBM and SXM sockets, meaning you can still run a bonkers GPU setup with standard PCIe gear and stay (just about) under 10kW per chassis.

Nvidia's RTX Pro 6000 Blackwell GPU, unveiled at GTC 2025, now appears in online listings with prices ranging from $8,565 to over $11,000, depending on the region and retailer. In Japan, a retailer has listed the GPU for ¥ $ 1,630,600 (approximately $11,326), while a UK-based retailer, Scan, accepts pre-orders at £7,859.99 (around $10,433).
In the US, enterprise-focused retailer Connection has listed the RTX Pro 6000 Blackwell at $8,565, marking a 26 per cent increase over the previous-generation RTX 6000 Ada. Some retailers, such as Tech-America and Directdial, offer pre-orders at a reduced price of $7,673.