NVIDIA RTX PRO 6000 Unlocks GB202’s Full Potential, Surpasses GeForce RTX 5090 in Gaming Benchmarks

NVIDIA’s new RTX PRO 6000 has emerged as the most powerful graphics card on the market—not just for workstations, but surprisingly, for gaming as well. Leveraging the full capabilities of the GB202 die based on the Blackwell architecture, this workstation GPU convincingly outperforms the flagship GeForce RTX 5090, previously touted as NVIDIA’s fastest gaming solution.

RTX PRO 6000 vs RTX 5090: A 10–15% Performance Lead in 4K Benchmarks

In a detailed comparison shared by renowned overclocker and tech reviewer Der8auer, the NVIDIA RTX PRO 6000 demonstrated a significant lead over the GeForce RTX 5090 across several synthetic and gaming benchmarks at 4K resolution. 

Powered by 24,064 CUDA cores—an increase of roughly 10% over the RTX 5090’s 21,760—the RTX PRO 6000 also features more RT cores, Tensor cores, and ROPs, fully unlocking the GB202’s capabilities. While designed as a professional-grade GPU with 96GB of GDDR7 VRAM, its raw power shows clear dominance in gaming scenarios as well.

Benchmark Highlights: FPS and Power Efficiency

In 3DMark Time Spy Extreme GT1, the RTX PRO 6000 achieved 176.5 FPS, surpassing the RTX 5090’s 156.1 FPS by 13%. In 3DMark Speedway, it continued its lead with an 8% performance advantage. The results are notable not only for their higher frame rates but also for the accompanying power draw and efficiency gains.

• In Cyberpunk 2077 (4K), the RTX PRO 6000 showed a 14% performance gain over the RTX 5090, albeit with 15% more power draw.

• In Remnant 2, the RTX PRO 6000 delivered 11% higher average FPS and an impressive 17% increase in 1% low FPS, with just 5% more power consumption.

Despite these results, the power consumption between tests varied—sometimes showing greater efficiency, other times drawing significantly more power to push performance boundaries.

Price and Market Position

The RTX PRO 6000 is clearly not positioned for mainstream gamers. With a staggering price of $10,000, it’s targeted at enterprise, workstation, and AI applications rather than traditional gaming use. Its massive 96GB of GDDR7 memory further underscores this focus. However, the fact remains: in terms of raw performance, it is currently the world’s fastest GPU for gaming workloads, outperforming the RTX 5090, even if it is over five times more expensive.

Real-World Takeaway

While gamers may not be rushing to replace their RTX 5090s with a workstation card anytime soon, the RTX PRO 6000’s performance underscores the headroom left in the GB202 architecture. It also demonstrates what’s possible when NVIDIA removes power or segmentation limitations typically present in consumer GPUs.

As NVIDIA continues refining Blackwell-based products, it’s clear that the architecture has far more potential than what’s currently available in the consumer RTX 5090 SKU.

Do you think NVIDIA should release a “Titan” or fully unlocked gaming version of the GB202 chip for enthusiasts? Let us know your thoughts.