datacenter
Thuê Nvidia H100.
Hopper
80GB VRAM
700W
From $0.80/hr
Per hour
$0.80
Per day
$19.20
Per week
$134.40
Per month
$576
Provider spread
15 providers ·
up to 76% cheaper at the low end
Cheapest · $0.80/hr on io.net
Median $2.20/hr
Most expensive · $3.29/hr on Lambda Labs
Price history
Daily median across providers.
Loading...
Where to rent it
All providers carrying this GPU.
| Provider | $/hr | 30d trend | |
|---|---|---|---|
| $0.80/hr | · | Launch ↗ | |
| $1.05/hr | ↑ 3% | Launch ↗ | |
| $1.25/hr | · | Launch ↗ | |
| $1.47/hr | · | Launch ↗ | |
| $1.59/hr | · | Launch ↗ | |
| $1.79/hr | · | Launch ↗ | |
| $1.85/hr | ↓ 23% | Launch ↗ | |
| $2.20/hr | · | Launch ↗ | |
|
V
Vultr
|
$2.30/hr | ↓ 42% | Launch ↗ |
| $2.40/hr | · | Launch ↗ | |
| $2.40/hr | · | Launch ↗ | |
| $2.49/hr | · | Launch ↗ | |
| $2.59/hr | · | Launch ↗ | |
| $2.99/hr | ↓ 39% | Launch ↗ | |
| $3.29/hr | ↓ 10% | Launch ↗ |
Buy vs rent
Should you rent or own?
Your usage
Rent on cloud
Per day
—
Per month
—
at $0.80/hr cheapest provider rate
Own on-prem
Electricity per day
—
700W TDP
Rent breakeven on $30,000 MSRP
— days
· — months
Workloads
Suitable workloads.
AI models that fit
See all 58 →
Run these on this GPU.
- Llama 3.1 405B 8× · fp16
- Kimi K2 8× · fp8
- DeepSeek V3 8× · fp8
Cloud instances
See all 14 →
Hyperscaler bundles.
Pre-configured on 14 clouds — from $1.47/hr total
($1.47/hr per GPU).
FAQ
Frequently asked.
How is the Nvidia H100 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 H100 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 H100?
Depends on the model size. With 80GB of VRAM you can fine-tune 70B models with QLoRA, or fully fine-tune 13B-class models.
Spot vs on-demand on the Nvidia H100 — 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 H100?
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 H100 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
$1.47/hr
Lowest $/hr per GPU
$1.47/hr
Providers
14
Instance shapes
14
| Provider | Instance | GPUs | vCPU | RAM | Disk | $/hr | $/hr per GPU | |
|---|---|---|---|---|---|---|---|---|
| tnr-h100-on-demand | 1× | — | — | — | $1.47/hr | $1.47/hr | ||
| rcrate-h100 | 1× | — | — | — | $1.79/hr | $1.79/hr | ||
| h100-on-demand | 1× | — | — | — | $2.20/hr | $2.20/hr | ||
|
V
Vultr
|
vc2-gpu-h100 | 1× | — | — | — | $2.30/hr | $2.30/hr | |
| h100-80g-pcie-spot | 1× | — | — | — | $2.40/hr | $2.40/hr | ||
| h100-sxm5 | 1× | — | — | — | $2.40/hr | $2.40/hr | ||
| gpu_1x_h100_pcie | 1× | — | — | — | $2.49/hr | $2.49/hr | ||
| nova-h100-on-demand | 1× | — | — | — | $2.59/hr | $2.59/hr | ||
| gpu-h100x1-80gb | 1× | — | — | — | $2.99/hr | $2.99/hr | ||
| g4.metal.h100-8x | 8× | — | — | — | $26.96/hr | $3.37/hr | ||
| BM.GPU.H100.8 | 8× | — | — | — | $80.00/hr | $10.00/hr | ||
| a3-highgpu-8g | 8× | — | — | — | $88.48/hr | $11.06/hr | ||
| p5.48xlarge | 8× | — | — | — | $98.32/hr | $12.29/hr | ||
| ND96isr_H100_v5 | 8× | — | — | — | $106.88/hr | $13.36/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 H100 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 H100 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 H100?
Depends on the model size. With 80GB of VRAM you can fine-tune 70B models with QLoRA, or fully fine-tune 13B-class models.
Spot vs on-demand on the Nvidia H100 — 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 H100?
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 H100 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
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
FP16
INT4
INT4
INT4
INT4
INT4
INT4
INT4
INT4
INT4
INT4
INT4
INT4
INT4
Models that run on this GPU.
GPU-count + quantization recommendations covering fine-tuning, inference, and run-it-yourself scenarios on the Nvidia H100.
Llama 3.1 405B
by Meta AI
Required GPUs
8× Nvidia H100
Total VRAM
640 GB
Parameters
405B
Context
128K tokens
Standard datacenter setup
Kimi K2
by Moonshot AI
Required GPUs
8× Nvidia H100
Total VRAM
640 GB
Parameters
1000B (32B active)
Context
256K tokens
MoE — only active experts loaded per token
DeepSeek V3
by DeepSeek
Required GPUs
8× Nvidia H100
Total VRAM
640 GB
Parameters
671B (37B active)
Context
128K tokens
MoE deployment, ~37B active params
DeepSeek R1
by DeepSeek
Required GPUs
8× Nvidia H100
Total VRAM
640 GB
Parameters
671B (37B active)
Context
128K tokens
DeepSeek R1 Distill Llama 70B
by DeepSeek
Required GPUs
2× Nvidia H100
Total VRAM
160 GB
Parameters
70B
Context
128K tokens
DeepSeek R1 Distill Qwen 32B
by DeepSeek
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
33B
Context
128K tokens
Llama 3.1 70B
by Meta AI
Required GPUs
2× Nvidia H100
Total VRAM
160 GB
Parameters
70B
Context
128K tokens
Llama 3.3 70B
by Meta AI
Required GPUs
2× Nvidia H100
Total VRAM
160 GB
Parameters
70B
Context
128K tokens
Mixtral 8x22B
by Mistral AI
Required GPUs
4× Nvidia H100
Total VRAM
320 GB
Parameters
141B (39B active)
Context
66K tokens
Qwen 2.5 72B
by Alibaba (Qwen Team)
Required GPUs
2× Nvidia H100
Total VRAM
160 GB
Parameters
73B
Context
128K tokens
Gemma 3 27B
by Google DeepMind
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
27B (27B active)
Context
128K tokens
Gemma 2 27B
by Google DeepMind
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
27B
Context
8K tokens
Qwen: Qwen3.6 27B
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
27B
Context
262K tokens
Qwen: Qwen3.5-27B
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
27B
Context
262K tokens
NVIDIA: Nemotron 3 Nano 30B A3B
by Nvidia
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
30B
Context
262K tokens
AllenAI: Olmo 3 32B Think
by Allen Institute for AI (AI2)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
32B
Context
66K tokens
Qwen: Qwen3 VL 32B Instruct
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
32B
Context
262K tokens
Qwen: Qwen3 VL 30B A3B Thinking
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
30B
Context
131K tokens
Qwen: Qwen3 VL 30B A3B Instruct
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
30B
Context
262K tokens
Tongyi DeepResearch 30B A3B
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
30B
Context
131K tokens
Qwen: Qwen3 30B A3B Thinking 2507
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
30B
Context
131K tokens
Qwen: Qwen3 Coder 30B A3B Instruct
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
30B
Context
160K tokens
Qwen: Qwen3 30B A3B Instruct 2507
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
30B
Context
262K tokens
Z.ai: GLM 4 32B
by Zhipu AI
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
32B
Context
128K tokens
Qwen: Qwen3 30B A3B
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
30B
Context
131K tokens
Qwen: Qwen3 32B
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
32B
Context
131K tokens
GLM-5.1
by Zhipu AI
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
32B
Context
203K tokens
ByteDance Seed: Seed-2.0-Lite
by Bytedance Seed
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
32B
Context
262K tokens
Baidu: ERNIE 4.5 VL 28B A3B
by Baidu
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
28B
Context
131K tokens
Qwen2.5 32B
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
32B
Context
131K tokens
Nemotron 3 Nano Omni 30B A3b Reasoning Fp8
by Nvidia
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
30B
Context
131K tokens
Cogito V1 Preview Qwen 32B
by Deepcogito
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
32B
Context
131K tokens
Gemma 3 27B It
by Google DeepMind
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
27B
Context
66K tokens
Qwen QwQ-32B
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
32B
Context
131K tokens
Qwen 2.5 Coder 32B Instruct
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
32B
Context
16K tokens
Qwen3 30B A3b Base
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
30B
Context
33K tokens
Gemma 3 27B Pt
by Google DeepMind
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
27B
Qwen3 30B A3B Instruct 2507 Lora
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
30B
Context
262K tokens
Nvidia Nemotron 3 Nano 30B A3b Bf16
by Nvidia
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
30B
Context
262K tokens
Medgemma 27B Text It
by Google DeepMind
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
27B
Context
131K tokens
Qwen2.5 32B Instruct
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
32B
Context
33K tokens
Gemma 3 27B It Lora
by Google DeepMind
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
27B
Gemma 4 31B It Lora
by Google DeepMind
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
31B
Context
262K tokens
Nemotron-3-Nano-Omni-30B-A3B-Reasoning
by Nvidia
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
30B
Context
262K tokens
gemma-4-31B-it-turbo
by Google DeepMind
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
31B
Context
262K tokens
DeepSeek R1 Distill Llama 70B
by DeepSeek
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
70B
Context
128K tokens
Single H100 with INT4 — sweet spot.
Llama 3.1 70B
by Meta AI
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
70B
Context
128K tokens
Mistral Large 2
by Mistral AI
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
123B
Context
128K tokens
Command R+
by Cohere
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
104B
Context
128K tokens
GPT-OSS 120B
by OpenAI
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
120B (5B active)
Context
128K tokens
GLM-4.5-Air
by Zhipu AI
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
106B (12B active)
Context
128K tokens
NVIDIA: Nemotron 3 Super
by Nvidia
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
120B
Context
1M tokens
Qwen: Qwen3.5-122B-A10B
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
122B
Context
262K tokens
Nvidia Nemotron 3 Super 120B A12b Fp8
by Nvidia
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
120B
Context
262K tokens
Nvidia Nemotron 3 Super 120B A12b Bf16
by Nvidia
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
120B
Context
262K tokens
Qwen3.5 122B A10b Fp8
by Alibaba (Qwen Team)
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
122B
Context
262K tokens
NVIDIA-Nemotron-3-Super-120B-A12B
by Nvidia
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
120B
Context
262K tokens
gpt-oss-120b-Turbo
by OpenAI
Required GPUs
1× Nvidia H100
Total VRAM
80 GB
Parameters
120B
Context
131K 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 H100 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 H100 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 H100?
Depends on the model size. With 80GB of VRAM you can fine-tune 70B models with QLoRA, or fully fine-tune 13B-class models.
Spot vs on-demand on the Nvidia H100 — 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 H100?
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 H100 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.