Rent H200 GPUs
on Vast.ai for
Based on the Hopper architecture, the NVIDIA H200 is a premium AI accelerator featuring 141GB of HBM3e memory and 4.8 TB/s bandwidth for enterprise-scale language model workloads.
No current offersThe rent link will appear when matching H200 listings are available.
H200-Ready Models
These popular AI models can run on the H200. Actual requirements vary by model, engine, precision, and configuration.
text
GLM 5.2
753B MoE model with 1M-token context for agentic reasoning, coding, and tool use
textvision
NVIDIA Nemotron 3.5 Super VL 120B A12B
Hybrid Mamba-attention MoE vision-language model with 12B active params, image and video input, reasoning control and tool calling
text
K2 Horizon MoVA 36B-A4B
Fully open MoE with Mixture-of-Values attention, 4B active parameters and 512K context
H200 GPU Details
| Category | H200 |
|---|---|
| GPU Name | GH100 |
| Architecture | Hopper |
| Process Size | 5 nm |
| Transistors | 80,000 million |
| Release Date | Nov 18th, 2024 |
| Base Clock | 1365 MHz |
| Boost Clock | 1785 MHz |
| Memory Size | 141 GB |
| Memory Type | HBM3e |
| Bandwidth | 3.36 TB/s |
| Tensor Cores | 528 |
| FP16 (half) | 241.3 TFLOPS (4:1) |
| FP32 (float) | 60.32 TFLOPS |