We are pleased to introduce the latest innovation from Nvidia: the RTX Pro 6000 Blackwell workstation graphics card. Based on the advanced Blackwell architecture, this card features a powerful GB202 graphics processor with an impressive 24,064 CUDA cores and 188 stream multiprocessors operating at speeds of up to 2,617 MHz. In comparison, the gaming flagship GeForce RTX 5090 also utilizes the GB202 chip but offers slightly fewer cores at 21,760, running at 2,410 MHz.

The RTX Pro 6000 is equipped with 96GB of GDDR7 memory, complete with ECC support and a substantial 1,792 GB/s memory bandwidth via a 512-bit bus, whereas the RTX 5090 features a 32GB memory but maintains the same bandwidth.

The graphics card is available in three distinct models to cater to diverse professional needs: Standard Workstation Edition, Max-Q Workstation Edition, and Server Edition. The Standard Workstation Edition utilizes a dual-fan, dual-slot open cooling system with a total graphics power draw of 600W, delivered through a single 16-pin PCIe 5.0 power connector. Meanwhile, the Max-Q Edition is optimized for intense workload environments, capped at 300W power usage and employing air-blown cooling. The Server Edition, ideal for data centers, relies on server fans for passive cooling. This card supports PCIe 5.0 for doubled bandwidth and is equipped with four DisplayPort 2.1b ports, facilitating multi-monitor setups.

Performance benchmarks using Geekbench 6 OpenCL reveal the RTX Pro 6000 scoring 368,219 points, slightly below the RTX 5090's 376,858, a variance of approximately 2.3 percent. Notably, in specialized tests, the RTX Pro 6000 excels past the RTX 5090 in activities such as Horizon Detection (21.3 GPixel/s), Edge Detection (32.1 GPixel/s), and Gaussian Blur (36.3 GPixel/s), although the RTX 5090 leads in Background Blur (263.9 FPS versus 310.7 FPS) and Face Detection (196.7 FPS versus 241.5 FPS). Testing constraints limited OpenCL memory access to 23.8GB, indicating the potential for performance improvements with more optimized drivers.

According to GameTechBench's full path tracking tests, the RTX Pro 6000 surpasses the RTX 5090 by roughly 5% in 4K resolution and holds a 2% edge in offline path tracking rendering. The card's substantial 96GB memory and 752 next-generation Tensor cores significantly enhance its capacity to manage large-scale 3D models, AI inference computations, and real-time ray tracing. The ninth-generation NVENC engine supports H.264 and HEVC encoding in 4:2:2, improving video encoding speeds and AV1 encoding quality by about 5%. The sixth-generation NVDEC engine doubles H.264 decoding throughput, supporting professional video editing applications.

The Blackwell architecture brings numerous technological breakthroughs, including integrated neural shaders within its streaming multiprocessors, embedding AI capabilities into programmable shaders, and the advancement of RTX Mega Geometry alongside DLSS 4.0 multi-frame generation technologies. With a single-precision floating-point performance reaching 125 TFLOPS and AI capacities extending to 4,000 TOPS, the card's dual-sided 16-module array of 3GB GDDR7 memory provides a total 96GB capacity, suited for VR, scientific computing, and multi-application workflows.

Priced between $8,435 and $8,565 at U.S. retailer Connection—approximately 26% above the RTX 6000 Ada—the RTX Pro 6000 caters to enterprise demands. Both the Standard Workstation Edition and Max-Q Edition share similar price points, with the Server Edition distributed through partners like Cisco and Dell. Availability is prompt for the Workstation Edition, while Server and Mobile Editions are expected in May and June respectively, integrated by brands such as Dell and HP. Expectations for the elite RTX Pro 6000 X Blackwell model include a GB202 chip with 192 stream multiprocessors.

At GDC 2025, the EmberGen real-time visual effects software exhibited a 40% performance boost on the RTX Pro 6000 over the RTX 4090. Render Network highlighted its remarkable efficiency in neural rendering and AI-driven creative processes during GTC 2025.

Geared towards professionals in game development, AI training, content creation, and industrial design, the RTX Pro 6000 is designed for users demanding large video memory and computational power. While initial assessments indicate minor performance discrepancies compared to the RTX 5090 due to driver constraints, future driver optimizations will likely unveil more of its potential. Its high energy consumption and premium pricing align it with enterprise-level use, differentiating it from consumer-focused offerings through the RTX Pro branding to refine market positioning. Ongoing evaluations will further reveal its true capabilities.