Lenovo ThinkStation PGX

Key features of the Lenovo ThinkStation PGX mini PC

Data centre performance in a miniature body

At the heart of the Lenovo ThinkStation PGX is the GB10 Grace Blackwell superchip. This unique combination features a 20-core ARM processor (10× Cortex-X925 + 10× Cortex-A725) and a Blackwell architecture GPU with 5th generation Tensor cores and 4th generation RT cores. Both parts of the superchip are connected by NVLink-C2C technology, which provides 5× the throughput of PCIe 5th generation and allows the CPU and GPU to share a single 128GB pool of unified LPDDR5X memory. The result is tremendous performance that consumes a maximum of 240W and fits on any desk. This shared memory, without the traditional separation of GPU VRAM and system RAM, is the key advantage of the Lenovo ThinkStation PGX over standard desktop builds. Where a GeForce RTX 5090 graphics card offers 32GB of memory, the Lenovo ThinkStation PGX makes four times that amount available for AI models, without needing to quantise or split models across multiple devices.

A personal supercomputer that fits on any desk

Lenovo ThinkStation PGX proves that you no longer need a server room or a rack full of hardware to run the largest AI models. With dimensions of 150 × 150 × 50.5mm and a weight of 1.2kg, the Lenovo ThinkStation PGX fits on any desk next to your laptop. Yet it offers performance that, just a few years ago, would have required an entire server filled with graphics cards. Four USB-C 4.0 ports with speeds of up to 40Gb/s, Wi-Fi 7, 10Gb Ethernet, and an HDMI output ensure all your accessories can be connected without needing an external hub or docking station. Lenovo ThinkStation PGX is designed for developers, scientists, and AI professionals who want the full power of a local, personal AI supercomputer without being dependent on the cloud.

Precision cooling for maximum thermal efficiency

Lenovo ThinkStation PGX handles even the most demanding AI tasks thanks to its sophisticated cooling solution. A pair of fans with seven-speed control ensures smooth and precise airflow according to the current load. The system delivers 1.6× more efficient cooling than comparable compact workstations. The result is a lower operating temperature and stable maximum performance, even under prolonged load.

Scalable architecture

A single Lenovo ThinkStation PGX can handle models with up to 200 billion parameters. Two linked AI PCs push the boundary much further. With integrated NVIDIA ConnectX-7 network technology, you can connect two Lenovo ThinkStation PGX systems into a single computational unit to handle models like Llama 3.1 with 405 billion parameters. Ultra-fast data throughput between nodes with minimal latency is ideal for large-scale distributed AI tasks, while hardware acceleration for TLS, IPsec, and MACsec ensures encrypted transmission without burdening the processor.

NVIDIA AI software ecosystem and task management

Lenovo ThinkStation PGX supports the complete NVIDIA AI software stack, which simplifies the training, management, and deployment of AI tasks in a local environment. The device lets you scale performance across your local infrastructure and integrate it into broader AI solutions without relying on cloud services.

Fine-tune AI models with up to 70 billion parameters

Use the 128GB of unified memory in the Lenovo ThinkStation PGX to fine-tune pre-trained models with up to 70 billion parameters right at your workstation. Training on your own data allows you to specialise AI models for specific needs, industry data, or particular use cases – without sending sensitive data to the cloud or paying for expensive GPU instances. The result is a model tailored precisely to your scenario, created locally, securely, and under your full control.

Lenovo ThinkStation PGX delivers top-tier data science

The combination of 128GB of unified memory and 1 PFLOP of performance in parallel throughput makes the Lenovo ThinkStation PGX the ideal workstation for demanding data analytics and machine learning projects. Extensive datasets, complex computational models, and intensive training and analysis processes that once required a powerful cloud cluster or dedicated server can now run directly on your desk. Fast, local, and without waiting for remote infrastructure.

Inference for models with up to 200 billion parameters

The 5th generation Tensor cores with FP4 format support achieve performance of up to 1 PFLOP. Combined with 128GB of system memory, this enables you to run inference on state-of-the-art AI models with up to 200 billion parameters right on your desk. Test, verify, and deploy models like Llama, Gemma, or Qwen locally and in real time – without cloud latency, shared infrastructure, or the risk of sensitive data leaks.

Develop applications for robotics, industry, and smart cities

Lenovo ThinkStation PGX is an exceptional platform for developing robotic systems, smart city solutions, and computer vision applications. The pre-installed NVIDIA Isaac framework for robotics, Metropolis for video analytics, and Holoscan for real-time data processing allow developers to fully harness its power and rapidly prototype and deploy applications – locally, without dependence on cloud infrastructure, and with full support from the NVIDIA ecosystem.