Acer has announced the Veriton GN100 AI Mini Workstation, a space-efficient personal AI computing device designed for local large language model inference and scalable AI workloads. The Veriton GN100 features the NVIDIA GB10 Grace Blackwell Superchip, delivering up to 1 petaFLOP (PFLOPS) of FP4 AI performance, and is built with 128 GB of LPDDR5x unified system memory alongside 4 TB of encrypted NVMe M.2 SSD storage.
The mini workstation is equipped with a 20-core Arm-based processor, next-generation CUDA cores, and fifth-generation Tensor Cores, providing server-grade performance in a compact 150 x 150 x 50.5 mm chassis with a weight of less than 1.5 kg. Security and manageability features include a Kensington lock slot and local AI model execution for enhanced data privacy.
For users requiring additional capacity, two Veriton GN100 units can be connected via an integrated NVIDIA ConnectX-7 Smart Network Interface Card (NIC), enabling support for AI models up to 405 billion parameters. The system supports Wi-Fi 7, Bluetooth 5.1 or above, four USB 3.2 Type-C ports, HDMI 2.1b output, and an RJ-45 Ethernet interface.
The Veriton GN100 comes pre-installed with the DGX Base operating system and the latest NVIDIA AI Software Stack, supporting development frameworks and tools such as PyTorch, Jupyter, and Ollama for AI prototyping, fine-tuning, and deployment. While the solution targets local AI model execution, it can also integrate into larger accelerated data center or cloud environments as needed.
Acer reports that the Veriton GN100 will be available in North America starting at USD 3,999, in EMEA at EUR 3,999, and in Australia at AUD 6,499. Specifications, pricing, and availability may vary by region.
Source: Acer
