INFRA Signal 157
System76 launches Thelio Mira AI workstation with dual RTX Pro 6000 GPUs and 192 GB GPU memory
Thelio Mira AI workstation provides up to 192 GB of GPU memory via two NVIDIA RTX Pro 6000 Blackwell GPUs, targeting local AI development.
Engineers can train and fine-tune larger AI models on a single workstation without relying on external cloud resources. The combination of high GPU memory, liquid-cooled Blackwell GPUs, and a choice of recent Linux distributions supports sustained workloads for tasks such as computer vision, simulation, and inference. Open-source hardware design and local manufacturing in Denver also simplify supply chain and support considerations.
Written by elseif from the cluster below · every claim links back to a sourceThe three things worth knowing
Dual NVIDIA RTX Pro 6000 Blackwell GPUs deliver up to 192 GB of GPU memory for AI workloads.
The workstation pairs these GPUs with up to a 16-core AMD Ryzen 9000 Series CPU and 192 GB DDR5 RAM.
It ships with Pop!_OS 24.04 LTS (COSMIC) or Ubuntu options, includes liquid cooling, and is built as open-source hardware in the USA.
THE READ
What the cluster adds up to.
Thelio Mira AI workstation introduces a dual-GPU configuration that provides 192 GB of GPU memory, a specification aimed at local AI development. This configuration uses two NVIDIA RTX Pro 6000 Blackwell graphics cards, which are liquid-cooled to sustain performance. By integrating this memory in a single chassis, the workstation reduces the need for multi-node setups for many model sizes.
Beyond the GPUs, the system can be configured with up to a 16-core AMD Ryzen 9000 Series processor and 192 GB of DDR5 RAM, offering balanced compute and memory capacity. Storage options include multiple M.2 PCIe Gen5 and Gen4 NVMe slots plus SATA bays, while power is supplied by either a 750W or 1000W SFX 80+ Gold unit depending on the GPU configuration. The chassis measures 17.31×9.96×15.12 inches and features extensive front and rear I/O, including USB-C, USB-A, and dual 5GbE LAN ports.
For engineers, the workstation enables tasks listed in its feature set: multi-GPU AI training, model fine-tuning, computer vision, image generation, simulations, and inference, all runnable on local hardware. The availability of recent Linux distributions, Pop!_OS 24.04 LTS with COSMIC, Ubuntu 24.04 LTS, or Ubuntu 26.04 LTS, provides a familiar software stack for AI frameworks. Liquid cooling and open-source hardware design aim to maintain reliability and ease of maintenance.
Limitations stem from the fixed dual-GPU layout; scaling beyond 192 GB of GPU memory requires external solutions or additional chassis. The workstation’s power draw (up to 1000W) and physical size may constrain deployment in space-limited racks. Warranty coverage is limited to 1-, 2-, or 3-year options, and international orders incur separate taxes and shipping costs, which engineers must factor into procurement planning.
Written by elseif from the cluster below · checked for specifics the sources never containedTHE CLUSTER
↗