PNY NVIDIA RTX PRO 6000 Blackwell 96GB GDDR7 workstation
The 96GB GDDR7 memory on a 512-bit bus delivers 1790 GB/s bandwidth, paired with 24064 CUDA cores and 126 FP32 TFLOPS for heavy compute workloads. Its dual-fan design fits a 267mm, dual-slot card, and the 4x DisplayPort 2.1b outputs support 7680 x 4320 resolution, making it a practical choice for high-density workstations. Best for AI researchers and engineers training large models or running real-time simulations, where the 95.5/100 AI LLM score and ECC memory justify the 600W TDP and 1000W PSU requirement.
Sem preço atual de lojas
Resumo
The 30-Second Version
The PNY RTX PRO 6000 Blackwell is the most powerful workstation card we've ever tested, topping our charts in every performance category. It's also $16,000 and pulls 600W, so it's strictly for enterprise buyers. If you need 96GB of VRAM for AI or rendering, this is the card. If not, look elsewhere.
Pros & Cons
Prós
- 96GB of GDDR7 means you can load massive AI models without breaking a sweat 100th
- Top of our database in raster, compute, AI, VRAM, and ray tracing 100th
- ECC memory keeps workstation workloads stable during long renders 100th
- Dual-slot design fits in more workstations than you'd expect for this class 100th
Contras
- 600W TDP will spike your power bill and heat up your office
- At $16,000, it's priced for enterprises, not enthusiasts
- Thermals are middle of the pack for such a power-hungry card
- Connectivity is just above average, not the best we've seen
As provas
Performance
The numbers here are stupid good. 126 FP32 TFLOPS, 96GB of GDDR7 on a 512-bit bus pushing 1,790 GB/s, and 24,064 CUDA cores. It tops our charts in every performance category we track. AI workloads? Best-in-class. Compute? Top of the heap. Ray tracing? Also number one. The 2,617MHz boost clock isn't the highest we've seen, but raw clock speed isn't the story here. It's the sheer scale. The only real knock is efficiency. At 600W TDP with a 1,000W PSU recommendation, this card drinks power like it's going out of style. You'll need serious cooling and a beefy power supply to keep it fed.
Specifications
Full Specifications
Chipset
| GPU Chipset | NVIDIA RTX PRO 6000 Blackwell |
| Brand | NVIDIA |
| Architecture | Blackwell |
| Class | workstation |
| Process | 5 |
Memory
| VRAM | 96 GB |
| Memory Type | GDDR7 |
| Memory Bus | 512 |
| Bandwidth | 1790 |
| ECC | Yes |
Compute
| CUDA Cores | 24064 |
| Tensor Cores | 752 |
| RT Cores | 188 |
| Base Clock | 1590 |
| Boost Clock | 2617 |
| FP16 TFLOPS | 126 |
| FP32 TFLOPS | 126 |
Performance
| PassMark G3D | 39287 |
| GPU Compute | 26392 |
Power & Cooling
| TDP | 600 |
| Recommended PSU | 1000 |
| Power Connectors | 1x 16-Pin PCI Connector |
| Cooler | Dual Fan |
| Fans | 2 |
Connectivity
| Interface | PCI Express 5.0 x16 |
| DisplayPort | 4 |
| HDMI | 0 |
| Max Displays | 4 |
| Max Resolution | 8K (4320p) |
Physical
| Length | 267 |
| Slot Width | 2 |
| Form Factor | Full Height |
| Weight | 2.3 kg / 5.1 lbs |
Features
| Ray Tracing | Yes |
| Upscaling | DLSS 4 |
| AV1 Encode | Yes |
| VR Ready | Yes |
| DirectX | 12 |
| RGB | No |
vs Competition
Stacked against the ASUS TUF RTX 5090, the MSI Ventus 5070 Ti, or the Gigabyte AERO 5080, the RTX PRO 6000 is in a different league entirely. Those cards are great for gaming and light content work, but they top out at 32GB of VRAM. This thing has triple that. The Sapphire NITRO+ and ZOTAC SOLID are also solid cards, but they're not even playing the same sport. If you need 96GB of ECC memory and PCIe 5.0 bandwidth, there's no real competition here. You're buying the top of the mountain.
| Spec | PNY NVIDIA RTX PRO 6000 Blackwell | ASUS TUF Gaming TUF-RTX5090-32G-GAMING | Gigabyte AERO GV-N5080AERO OC-16GD | MSI Ventus RTX 5070 Ti 16G VENTUS 3X PZ OC | ZOTAC APOCALYPSE ZT-B50800C-10P | Sapphire NITRO+ 11322-01-40G |
|---|---|---|---|---|---|---|
| Chipset | NVIDIA RTX PRO 6000 Blackwell | GeForce RTX 5090 | GeForce RTX 5080 | GeForce RTX 5070 Ti | GeForce RTX 5080 | Radeon RX 7900 XTX |
| Vram | 96GB GDDR7 | 32GB GDDR7 | 16GB GDDR7 | 16GB GDDR7 | 16GB GDDR7 | 24GB GDDR6 |
| Boost Clock Mhz | 2617 | 2407 | 2730 | 2482 | 2730 | 2680 |
| Tdp W | 600 | 575 | 360 | 300 | 360 | 339 |
| Memory Bus Bits | 512 | 512 | 256 | 256 | 256 | 384 |
| Ray Tracing | true | true | true | true | true | true |
| GPU Class | workstation | consumer | consumer | consumer | consumer | consumer |
| Compare | Compare | Compare | Compare | Compare |
| Produto | AI | Rt | Vram | Raster | Compute | Feature | Thermal | Efficiency | Connectivity | Prova social | Affordability |
|---|---|---|---|---|---|---|---|---|---|---|---|
| PNY NVIDIA RTX PRO 6000 Blackwell | 99.8 | 99.8 | 99.8 | 99.9 | 97.5 | 99.9 | 43.4 | 22.1 | 87.1 | 0.9 | 1 |
| ASUS TUF Gaming TUF-RTX5090-32G-GAMING Compare | 96.9 | 90.1 | 97.6 | 98.7 | 97.6 | 96.2 | 72.4 | 24.6 | 99.3 | 99.1 | 3.5 |
| Gigabyte AERO GV-N5080AERO OC-16GD Compare | 85.3 | 95.2 | 88.7 | 94.2 | 93.2 | 96.2 | 91.5 | 57.8 | 94 | 88.8 | 15.1 |
| MSI Ventus RTX 5070 Ti 16G VENTUS 3X PZ OC Compare | 79.2 | 92.2 | 85.6 | 88.8 | 87.9 | 96.2 | 91.5 | 72.9 | 94 | 88.8 | 21 |
| ZOTAC APOCALYPSE ZT-B50800C-10P Compare | 85.3 | 97.5 | 88.7 | 94.2 | 93.2 | 96.2 | 91.5 | 57.8 | 94 | 99.1 | 19.1 |
| Sapphire NITRO+ 11322-01-40G Compare | 94.9 | 96.1 | 91.7 | 83.8 | 95.5 | 83.9 | 91.5 | 52.7 | 98.7 | 78.4 | 17.6 |
Preço
Value & Pricing
Let's be real. At $16,000, this card is not a value play. Our affordability score puts it dead last, and that's not changing anytime soon. But value isn't just about price. If you're training domain-specific AI models or rendering billion-polygon engineering designs, this card pays for itself in time saved. For the right buyer, it's not an expense. It's an investment. For everyone else, it's a very expensive way to play games at 8K.
Saiba mais
Overview
The PNY RTX PRO 6000 Blackwell is a workstation card that doesn't care about your gaming benchmarks. It's built for AI training, massive simulations, and rendering jobs that would make lesser cards cry. With 96GB of GDDR7 and 24,064 CUDA cores, this thing is an absolute monster.
It's also $16,000. So unless your company is footing the bill, you're probably just here to gawk. And honestly? Fair. This card sits at the top of our database in raster, compute, AI, VRAM, and ray tracing. It's the best of the best, and it knows it.
Common Questions
Q: Is this card good for gaming?
It'll run games fine, but that's like using a supercomputer to play solitaire. The RTX PRO 6000 is built for AI, rendering, and simulation workloads. A $2,000 RTX 5090 will give you nearly identical gaming performance for a fraction of the price.
Q: What power supply do I need?
NVIDIA recommends a 1,000W PSU minimum. The card itself pulls 600W, so you'll want headroom for the rest of your system. Make sure you have a 16-pin power connector available.
Q: Can I use this for AI model training?
Absolutely. With 96GB of GDDR7 and 752 Tensor cores, this card is built for exactly that. It's one of the best options on the market for training large language models and running inference on complex AI workloads.
Who Should Skip This
Skip this if you're a gamer, a streamer, or a casual content creator. You don't need 96GB of VRAM to play Cyberpunk, and you definitely don't need to spend $16,000 to edit YouTube videos. The 600W power draw also means you'll need a serious cooling setup. If your workloads fit in 16-32GB of VRAM, buy a 5080 or 5090 and spend the other $14,000 on literally anything else.
Verdict
This card is for AI researchers, VFX studios, and engineering firms that need maximum compute without compromise. If you're training large language models, running real-time physics simulations, or rendering at 8K, the RTX PRO 6000 is the best tool money can buy. If you're a gamer or a hobbyist, save your $16,000 and buy a 5090. You'll be happier, and your wallet will thank you.
Usage Scores
- Geral81.9
- AI/LLM96.3
- Budget42
- Gaming 4k64.2
- Ray Tracing68.4
- Workstation99.1
- Gaming 1080p59.3
- Gaming 1440p62.6
- Content Creation88