1× NVIDIA GH200 96 GB

Hardware

GPU

NVIDIA GH200 96 GB

GPU count

1

VRAM per GPU

96 GB advertised

Aggregate VRAM

96.0 GB advertised

System RAM

480 GB

CPU

Grace Neoverse V2

Storage

960 GB NVME

Network

15360 public

AI memory capability

Default catalog below is Q4 at 8K. Statuses reuse Phase 2 (1 GPU) or Phase 8 (2–4 identical GPUs). GPU counts above 4 are not modeled. Public wording is memory-capable, not serving throughput.

Memory-capable at Q4/8K: 54 published models

• Qwen2.5 0.5B Instruct
• Llama 3.2 1B Instruct
• Qwen2.5 1.5B Instruct
• Qwen2.5-Coder 1.5B Instruct
• SmolLM2 1.7B Instruct
• Gemma 2 2B Instruct
• Qwen2.5 3B Instruct
• Llama 3.2 3B Instruct
• Phi-3.5 Mini Instruct
• Phi-3 Mini 128K Instruct
• Phi-4 Mini Instruct
• Code Llama 7B Instruct
• StarCoder2 7B
• Falcon 7B Instruct
• OpenChat 3.5 7B
• Mistral 7B Instruct v0.3
• Falcon 3 7B Instruct
• DeepSeek R1 Distill Qwen 7B
• Qwen2.5 7B Instruct
• Qwen2.5-Coder 7B Instruct
• Qwen2 7B Instruct
• DeepSeek R1 Distill Llama 8B
• Hermes 3 Llama 3.1 8B
• Llama 3.1 8B Instruct

Serving performance

Measured vLLM serving results for this GPU identity (1× NVIDIA GH200 96 GB), not a specific rental host. Tensor parallel size is stored explicitly and is not assumed equal to GPU count. Request rate is separate from concurrency; this corpus is closed-loop / max-concurrency.

No measured vLLM serving rows map to this GPU count yet.

Current rental offers