NVIDIA A100 PCIe (40 GB HBM2e, 312 TFLOPS FP16, MIG support) · AMD EPYC 7763 (64c/128t) · 512 GB DDR4 ECC · 8× 3.84 TB NVMe SSD · 25 Gbps. Production AI workloads, large LLM training (70B+), multi-tenant inference via MIG. NVLink çoklu node opsiyonel.

NVIDIA RTX A5000 (24 GB GDDR6) · AMD Ryzen Threadripper PRO 5965WX (24c/48t) · 256 GB DDR4 ECC · 4× 1.92 TB NVMe SSD · 1 Gbps. LLM fine-tune (13B-30B), high-res Stable Diffusion XL, video AI rendering, embeddings training.

NVIDIA L40S (48 GB GDDR6, Ada Lovelace, 2026 nesil) · AMD EPYC 7543 (32c/64t) · 256 GB DDR4 ECC · 4× 3.84 TB NVMe SSD · 10 Gbps. 70B LLM inference, large diffusion models, multi-modal training. RTX A6000'in halefi.

NVIDIA H100 PCIe (80 GB HBM3, 3 PFLOPS FP8, Transformer Engine) · AMD EPYC Genoa (96c/192t) · 1 TB DDR5 ECC · 16× 3.84 TB NVMe Gen4 · 100 Gbps. Hyperscale LLM training (175B+), enterprise inference, çoklu H100 NVLink cluster opsiyonel. Teklif bazlı.

NVIDIA RTX A4000 (16 GB GDDR6, Ampere) · AMD Ryzen 9 5950X (16c/32t) · 128 GB DDR4 ECC · 2× 1.92 TB NVMe SSD · 1 Gbps. PyTorch, TensorFlow, JAX preinstalled. CUDA 12.x + cuDNN hazır. Stable Diffusion, LLM fine-tune (7B-13B), küçük model training için ideal.