MINIX has introduced its latest compact AI systems, the T4000 and T5000 GenAI mini workstations, expanding its portfolio in the growing edge AI segment. Announced on April 23, the new models are built on the NVIDIA Jetson Thor platform, pairing Arm Neoverse-V3AE CPUs with next-generation Blackwell GPU architecture to handle modern AI workloads in a small footprint.
The T4000 and T5000 are designed for local generative AI processing, offering up to 128GB of LPDDR5X unified memory. This configuration allows the systems to manage large language models in the 7B to 70B parameter range without relying on external cloud resources. MINIX reports that the systems can reach up to 2070 TFLOPS of FP4 sparse computing performance, which is aimed at improving inference speeds for developers and enterprise users working with AI models at the edge.
Both variants share an identical chassis, measuring 139.3 × 131 × 76.8 mm and weighing approximately 1420 grams. The enclosure combines metal and plastic materials and integrates a dual-turbine cooling setup to maintain consistent thermal performance during extended workloads. The compact design makes the systems suitable for deployment in environments where space is limited, such as small offices, labs, or embedded installations.
On the storage side, the mini workstations come pre-installed with a 1TB PCIe Gen4 SSD, offering fast data access for AI datasets and applications. Connectivity is a key focus, with dual 10GbE Ethernet ports included for high-speed networking. Wireless options consist of Wi-Fi 6E and Bluetooth 5.3, ensuring compatibility with modern devices and networks.
For display and peripheral support, the systems feature two HDMI 2.1 TMDS ports, along with four USB-A 5Gbps ports and one USB-C 10Gbps port. This range of I/O options allows users to connect multiple displays and accessories without the need for additional hubs.
With the T4000 and T5000, MINIX is targeting developers and businesses looking to run AI workloads locally. The combination of Jetson Thor hardware and Blackwell GPU architecture reflects a shift toward compact systems that can deliver high compute performance outside of traditional data centers.
