Demand for desktop AI compute among professional users is growing exponentially, from local large-model fine-tuning and inference to generative-AI design, data-science mining and 3D simulation modelling. Traditional desktop GPUs are bottlenecked by memory capacity, AI compute throughput and multi-task parallelism, and cannot fit the increasingly complex desktop AI-development and professional-visual-computing scenarios. NVIDIA RTX PRO desktop GPUs were built for exactly this.
As NVIDIA’s new flagship platform for desktop AI and professional visual computing, RTX PRO is built on the Blackwell architecture with fifth-generation Tensor Cores and fourth-generation RT Cores. Against the previous generation, AI inference throughput rises up to threefold and ray-tracing performance up to twofold. With RTX Mega Geometry neural graphics, the device can build physically precise, photoreal immersive 3D design scenes, supporting up to a hundredfold more ray-traced triangles. RTX PRO memory scales flexibly from 16 GB to 72 GB GDDR7 ECC, fitting AI model development, data-science research and digital simulation across workloads.
On 15 July at 14:00, Zhidx launched its “2026 Zhidx Open Class NVIDIA RTX PRO Session,” presented by Ding Qian, senior engineer at Ingram Micro China, on the topic “NVIDIA RTX PRO Blackwell Desktop GPU Accelerates AI and Professional Visual Computing.”
Editor’s note: This is an adapted translation of the original Zhidx report. It has been trimmed and restructured for readability for an international business audience. The full original (in Chinese) is at https://aiorang.com/c/MWM2YTc0MGZmMzYyODc2MWNiYzM=.
Translated and adapted from Zhidx (https://aiorang.com/c/MWM2YTc0MGZmMzYyODc2MWNiYzM=).