datacenter
Thuê Nvidia H200.
Hopper
141GB VRAM
700W
From $0.46/hr
Per hour
$0.46
Per day
$11.16
Per week
$78.12
Per month
$335
Provider spread
7 providers ·
up to 87% cheaper at the low end
Cheapest · $0.46/hr on RunPod
Median $2.40/hr
Most expensive · $3.50/hr on DigitalOcean
Price history
Daily median across providers.
Loading...
Where to rent it
All providers carrying this GPU.
Buy vs rent
Should you rent or own?
Your usage
Rent on cloud
Per day
—
Per month
—
at $0.46/hr cheapest provider rate
Own on-prem
Electricity per day
—
700W TDP
Rent breakeven on $40,000 MSRP
— days
· — months
Workloads
Suitable workloads.
AI models that fit
See all 17 →
Run these on this GPU.
- Llama 3.1 405B 4× · fp16
- Kimi K2 4× · fp8
- DeepSeek V3 4× · fp8
Cloud instances
See all 6 →
Hyperscaler bundles.
Pre-configured on 6 clouds — from $2.99/hr total
($2.99/hr per GPU).
FAQ
Frequently asked.
How is the Nvidia H200 price calculated?
We pull live listings from each provider's public API, take the median hourly rate across active offers, and refresh every hour. The rate shown is the median, so a single low-ball spot offer can't distort the headline.
Why does the Nvidia H200 cost different amounts on different providers?
P2P marketplaces like Vast.ai aggregate offers from individual hosts who set their own rates — bidding pushes prices down. First-party clouds (Lambda, hyperscalers) charge a managed-service premium for support, SLAs, and integrated networking. Decentralized networks (io.net, Akash) settle in tokens, which adds volatility but often the lowest base rate.
Can I really train an LLM on a single Nvidia H200?
Depends on the model size. With 141GB of VRAM you can fine-tune 70B models with QLoRA, or fully fine-tune 13B-class models.
Spot vs on-demand on the Nvidia H200 — which should I rent?
On-demand keeps the same instance until you stop it; spot (or interruptible) is cheaper but the host can reclaim it when a higher-paying job lands. Use on-demand for training runs and anything stateful. Use spot for stateless inference, batch jobs, and experiments where a checkpoint every few minutes is enough to recover.
How is hourly billing measured for the Nvidia H200?
Most providers bill per-second once the instance is running, with a small minimum (often 60 seconds). A handful of first-party clouds round up to the minute. Either way, headline $/hr is the right comparison unit.
Does the region of the host affect the Nvidia H200 price?
Yes — US and EU regions usually carry a premium over LATAM, India, and parts of APAC, especially on first-party clouds. P2P marketplaces hide this behind one global price because supply moves wherever bids exist.
Cloud instance options
Pre-configured instances on hyperscalers.
Whole-instance bundles (GPU + vCPU + RAM + disk) on the major clouds. Per-GPU rate often drops as the count rises. View = spec page · Launch = sign up (affiliate).
Cheapest bundle
$2.99/hr
Lowest $/hr per GPU
$2.99/hr
Providers
6
Instance shapes
6
| Provider | Instance | GPUs | vCPU | RAM | Disk | $/hr | $/hr per GPU | |
|---|---|---|---|---|---|---|---|---|
| h200-sxm5 | 1× | — | — | — | $2.99/hr | $2.99/hr | ||
| gpu_1x_h200 | 1× | — | — | — | $3.29/hr | $3.29/hr | ||
| gpu-h200x1-141gb | 1× | — | — | — | $3.50/hr | $3.50/hr | ||
| BM.GPU.H200.8 | 8× | — | — | — | $96.00/hr | $12.00/hr | ||
| a4x-highgpu-8g | 8× | — | — | — | $108.00/hr | $13.50/hr | ||
| ND96isr_H200_v5 | 8× | — | — | — | $118.40/hr | $14.80/hr |
Looking for the cheapest rate?
Hyperscaler bundles include managed networking + SLAs. Raw per-GPU rental on P2P marketplaces is typically 3–10× cheaper.
See raw rental rates on the Overview tab →
FAQ
Cloud instances — common questions.
How is the Nvidia H200 price calculated?
We pull live listings from each provider's public API, take the median hourly rate across active offers, and refresh every hour. The rate shown is the median, so a single low-ball spot offer can't distort the headline.
Why does the Nvidia H200 cost different amounts on different providers?
P2P marketplaces like Vast.ai aggregate offers from individual hosts who set their own rates — bidding pushes prices down. First-party clouds (Lambda, hyperscalers) charge a managed-service premium for support, SLAs, and integrated networking. Decentralized networks (io.net, Akash) settle in tokens, which adds volatility but often the lowest base rate.
Can I really train an LLM on a single Nvidia H200?
Depends on the model size. With 141GB of VRAM you can fine-tune 70B models with QLoRA, or fully fine-tune 13B-class models.
Spot vs on-demand on the Nvidia H200 — which should I rent?
On-demand keeps the same instance until you stop it; spot (or interruptible) is cheaper but the host can reclaim it when a higher-paying job lands. Use on-demand for training runs and anything stateful. Use spot for stateless inference, batch jobs, and experiments where a checkpoint every few minutes is enough to recover.
How is hourly billing measured for the Nvidia H200?
Most providers bill per-second once the instance is running, with a small minimum (often 60 seconds). A handful of first-party clouds round up to the minute. Either way, headline $/hr is the right comparison unit.
Does the region of the host affect the Nvidia H200 price?
Yes — US and EU regions usually carry a premium over LATAM, India, and parts of APAC, especially on first-party clouds. P2P marketplaces hide this behind one global price because supply moves wherever bids exist.
AI models
FP16
FP8
FP8
FP8
FP16
FP16
FP16
FP8
FP8
FP8
FP8
FP8
FP8
FP8
FP8
INT4
INT4
Models that run on this GPU.
GPU-count + quantization recommendations covering fine-tuning, inference, and run-it-yourself scenarios on the Nvidia H200.
Llama 3.1 405B
by Meta AI
Required GPUs
4× Nvidia H200
Total VRAM
564 GB
Parameters
405B
Context
128K tokens
Kimi K2
by Moonshot AI
Required GPUs
4× Nvidia H200
Total VRAM
564 GB
Parameters
1000B (32B active)
Context
256K tokens
DeepSeek V3
by DeepSeek
Required GPUs
4× Nvidia H200
Total VRAM
564 GB
Parameters
671B (37B active)
Context
128K tokens
DeepSeek R1
by DeepSeek
Required GPUs
4× Nvidia H200
Total VRAM
564 GB
Parameters
671B (37B active)
Context
128K tokens
NVIDIA: Llama 3.3 Nemotron Super 49B V1.5
by Nvidia
Required GPUs
1× Nvidia H200
Total VRAM
141 GB
Parameters
49B
Context
131K tokens
nim/nvidia/llama-3.3-nemotron-super-49b-v1
by Nvidia
Required GPUs
1× Nvidia H200
Total VRAM
141 GB
Parameters
49B
Context
16K tokens
Llama 3.3 70B
by Meta AI
Required GPUs
1× Nvidia H200
Total VRAM
141 GB
Parameters
70B
Context
128K tokens
Command R+
by Cohere
Required GPUs
1× Nvidia H200
Total VRAM
141 GB
Parameters
104B
Context
128K tokens
Llama 3.2 90B Vision
by Meta AI
Required GPUs
1× Nvidia H200
Total VRAM
141 GB
Parameters
90B
Context
128K tokens
GLM-4.5-Air
by Zhipu AI
Required GPUs
1× Nvidia H200
Total VRAM
141 GB
Parameters
106B (12B active)
Context
128K tokens
Qwen: Qwen3 Next 80B A3B Thinking
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H200
Total VRAM
141 GB
Parameters
80B
Context
262K tokens
Qwen: Qwen3 Next 80B A3B Instruct
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H200
Total VRAM
141 GB
Parameters
80B
Context
262K tokens
Qwen3 Next 80B A3b Instruct Fp8
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H200
Total VRAM
141 GB
Parameters
80B
nim/meta/llama-3.2-90b-vision-instruct
by Meta AI
Required GPUs
1× Nvidia H200
Total VRAM
141 GB
Parameters
90B
Context
16K tokens
Tencent: Hunyuan A13B Instruct
by Tencent
Required GPUs
1× Nvidia H200
Total VRAM
141 GB
Parameters
80B
Context
131K tokens
ByteDance Seed: Seed 1.6
by Bytedance Seed
Required GPUs
1× Nvidia H200
Total VRAM
141 GB
Parameters
200B
Context
262K tokens
Seed-1.8
by Bytedance
Required GPUs
1× Nvidia H200
Total VRAM
141 GB
Parameters
200B
Context
256K tokens
Renting for inference?
Pair these models with the cheapest provider on the rental table.
See rental rates on the Overview tab →
FAQ
AI models on this GPU.
How is the Nvidia H200 price calculated?
We pull live listings from each provider's public API, take the median hourly rate across active offers, and refresh every hour. The rate shown is the median, so a single low-ball spot offer can't distort the headline.
Why does the Nvidia H200 cost different amounts on different providers?
P2P marketplaces like Vast.ai aggregate offers from individual hosts who set their own rates — bidding pushes prices down. First-party clouds (Lambda, hyperscalers) charge a managed-service premium for support, SLAs, and integrated networking. Decentralized networks (io.net, Akash) settle in tokens, which adds volatility but often the lowest base rate.
Can I really train an LLM on a single Nvidia H200?
Depends on the model size. With 141GB of VRAM you can fine-tune 70B models with QLoRA, or fully fine-tune 13B-class models.
Spot vs on-demand on the Nvidia H200 — which should I rent?
On-demand keeps the same instance until you stop it; spot (or interruptible) is cheaper but the host can reclaim it when a higher-paying job lands. Use on-demand for training runs and anything stateful. Use spot for stateless inference, batch jobs, and experiments where a checkpoint every few minutes is enough to recover.
How is hourly billing measured for the Nvidia H200?
Most providers bill per-second once the instance is running, with a small minimum (often 60 seconds). A handful of first-party clouds round up to the minute. Either way, headline $/hr is the right comparison unit.
Does the region of the host affect the Nvidia H200 price?
Yes — US and EU regions usually carry a premium over LATAM, India, and parts of APAC, especially on first-party clouds. P2P marketplaces hide this behind one global price because supply moves wherever bids exist.