NVIDIA GB200 GPU: Specs, VRAM, Price & Benchmarks (2026)
The GB200 is a real GPU (launched 2024), but no provider on Aquanode is listing it for rental right now, so there is no live hourly price to quote. Availability changes as providers add and retire hardware; the specs, VRAM and price context below still apply.
How much VRAM does the GB200 have?
The GB200 has 384GB HBM3e per Superchip (2x Blackwell GPUs, 192GB each), with 16 TB/s aggregate (per Superchip) of peak memory bandwidth.
What can the GB200 run?
Popular open models from small to frontier scale, with the memory each needs and how many GB200 cards (384GB HBM3e per Superchip (2x Blackwell GPUs, 192GB each) each) that takes.
| Model | As published | FP8 | INT4 |
|---|---|---|---|
| Qwen/Qwen3-8B 8.2B | BF16: not supported | FP8: not supported | INT4: not supported |
| Qwen/Qwen2.5-14B-Instruct 14.8B | BF16: not supported | FP8: not supported | INT4: not supported |
| Qwen/Qwen3-32B 32.8B | BF16: not supported | FP8: not supported | INT4: not supported |
| Qwen/Qwen-72B 72.3B | BF16: not supported | FP8: not supported | INT4: not supported |
| MiniMaxAI/MiniMax-M2.7 228.7B | FP8: not supported | – | INT4: not supported |
| deepseek-ai/DeepSeek-R1 684.5B | FP8: not supported | – | INT4: not supported |
Estimates: weights at the stated precision plus a flat 20% for KV cache and overhead, at a moderate context length. A dash means the precision is not offered for that model (it is already published at that size). INT4 needs a published quantized checkpoint. Open any model for a per-GPU breakdown.
GB200 specs
| Architecture | launched 2024 |
| VRAM | 384GB HBM3e per Superchip (2x Blackwell GPUs, 192GB each) |
| Memory bandwidth | 16 TB/s aggregate (per Superchip) |
| Interconnect | NVLink-C2C, 900 GB/s CPU-GPU; NVLink 5 GPU-GPU within the NVL72 rack |
| TDP | 1,200W per Superchip module (2 GPUs + 1 Grace CPU) |
| Form factor | Superchip module (rack-scale, liquid-cooled; deployed in the GB200 NVL72 rack) |
Specs sourced from the vendor's public product page. See the source.
GPU Glossary: What is VRAM?, HBM, Tensor Cores, CUDA Cores, TFLOPS, NVLink vs PCIe
Good for
A rack-scale training and inference system, not a single GPU: one GB200 NVL72 rack packs 72 Blackwell GPUs and 36 Grace CPUs into one NVLink domain with 13.4TB of aggregate GPU memory, built for frontier-scale model training and the largest inference deployments. Aquanode does not list per-GPU GB200 rental today; see the live H100/H200/B200 pricing above for the same Blackwell/Hopper compute by the GPU-hour.
Not good for
Not a fit for anything short of a full-rack deployment: it is sold and deployed as a complete liquid-cooled system, not a card you rent or buy individually, and no cloud marketplace lists it at per-GPU granularity the way it lists H100 or B200.
Related guides
Other models in the same generation. The full list is in the GPU index.
Get notified when the price drops
GPU supply moves hourly. Tell us what you're waiting for and we'll email you when a matching offer appears across any provider we track.
