ABS AI Workstations — Precision Engineered for AI Creators

Which workstation is best for LLM training?

The ABS Zaurion Pro and Duo Ruby offer multi-GPU performance ideal for training and fine-tuning large language models. The Zaurion Pro supports up to 4× RTX 6000 Ada Generation GPUs, while the Duo Ruby provides dual RTX Pro 6000 Blackwell MaxQ GPUs for high-throughput AI training workloads.

How many GPUs do I need for AI workloads?

Most AI training tasks benefit from 2–4 GPUs, while inference-focused workloads often run effectively on a single GPU. For large language model training, we recommend the Zaurion Pro (4 GPUs) or Duo Aqua/Duo Ruby (2 GPUs). For inference and development, single-GPU models like Zaurion Aqua or Ruby are typically sufficient.

Can ABS workstations run 4× RTX GPUs?

Yes. The Zaurion Pro is specifically designed to support up to four NVIDIA RTX 6000 Ada Generation GPUs for high-throughput AI workloads, LLM training, and enterprise AI/ML applications. It features up to 2TB DDR5 ECC memory and PHISON aiDAPTIV+ AI optimization software.

What's the difference between RTX 6000 Blackwell and Ada Generation?

RTX 6000 Blackwell is NVIDIA's latest professional GPU architecture, offering up to 2.3× faster rendering and improved AI training performance compared to Ada Generation. Blackwell GPUs feature 96GB GDDR7 memory and are available in our Zaurion Aqua, Ruby, Duo Aqua, and Duo Ruby models. Ada Generation GPUs (available in Zaurion Pro) provide excellent performance for multi-GPU enterprise deployments.

Do ABS workstations come with AI software pre-installed?

Yes, optionally. ABS workstations can be configured with a pre-installed AI stack including PyTorch, TensorFlow, CUDA, RAPIDS, Triton, and Docker — all tuned and validated for workstation performance. This allows you to start training immediately without environment setup.