What's the difference between the H100 SXM, PCIe and NVL?

The H100 ships in 3 versions with different memory, bandwidth or interconnect. The figures at the top of this page are for the H100 SXM. H100 SXM: 80 GB, 3,350 GB/s, NVLink 900 GB/s H100 PCIe: 80 GB, 2,000 GB/s, NVLink 600 GB/s H100 NVL: 94 GB, 3,938 GB/s, NVLink 600 GB/s

Nvidia H100

Q: How much does the H100 cost per hour?

As of September 17, 2026, the median on-demand price is $3.38 per GPU per hour across 38 providers with a priced on-demand config. On-demand: 201 configs, median $3.38, cheapest $1.25 (in stock, Lium) Reserved: 176 configs, median $3.13, cheapest $1.53 (1 mo, in stock, HyperAI) Spot: 49 configs, median $2.00, cheapest $0.34 (in stock, Vast.ai) 6 providers also quote the H100 on a custom contract, priced per deal.

Q: Who has the cheapest H100?

The cheapest verified in-stock on-demand H100 tracked by GetDeploying is $1.25 per GPU per hour from Lium. Cheapest verified in stock on other billing types: reserved $1.53 (1 mo) from HyperAI and spot $0.34 from Vast.ai.

Q: How much does the H100 cost per month?

At 720 hours per month, one H100 costs an estimated $2,434 at the median on-demand price. Cheapest verified in stock: $900 per month on-demand, $1,102 reserved (1 mo), $245 spot.

Q: Where can I rent an H100?

GetDeploying currently tracks H100 configs from 53 providers. The cheapest verified in-stock on-demand configs come from Lium, Vast.ai, HyperAI and GPU.ai. See the full price comparison above for every provider and config.

Q: Are H100 prices going up or down?

As of September 17, 2026, the median on-demand price has risen about 7% over the past 90 days to $3.38 per GPU per hour, and about 11% over the past 12 months.

Q: How many H100 GPUs do I need?

One H100 runs models up to roughly 120B parameters at 4-bit quantization or 31B at 16-bit, assuming a 32K context. Larger models run across multiple GPUs: the model table above shows the estimated memory and GPU count for popular open-weight LLMs.

Q: Why choose the H100?

80GB HBM3 with 900 GB/s NVLink for multi-node training. FP8 Transformer Engine for improved training throughput over A100. The standard choice for large-scale AI training and high-throughput inference.

Q: When is the H100 not a good fit?

May be more than needed for single-GPU inference on models under 30B parameters. Consider the L40S or A10 for cost-effective inference workloads.

The standard data center GPU for large-scale AI training and inference.

Arch: Hopper
Memory: 80 GB HBM3 SXM
Bandwidth: 3.35 TB/s SXM
Released: Q3 2022

Compare with another GPU

Price history

Weekly median price per GPU per hour

Last 12 months

Aggregating historical prices...

At a glance Updated 7 minutes ago

Cheapest in stock: $1.25 /GPU/hr; on-demand · 1x PCIe · Lium
Median price (current): $3.38 /GPU/hr; on-demand
90-day trend: +7%; median on-demand
Coverage: 53 providers; 434 price points

Pricing & availability

How this list works

The two views

By provider shows one card per company, placed where its best offer ranked. By configuration lists every offer.

Order

What you can rent comes first: in stock, then waitlist, then not reported, then out of stock. Priced offers rank ahead of quote-only, and on-demand ahead of other billing types.

Within a group, five factors set the order:

Location: datacenter proximity, blended with provider HQ.
Price: hourly price, per GPU and in total.
Billing type: reserved ahead of spot, spot ahead of quote-only.
Specs: more VRAM, vCPUs and RAM.
Provider diversity: a provider's repeat rows rank slightly lower, so one company can't take all the top spots.

Search

We match the provider, its country, GPU model, form factor, billing type, availability and instance name. Matching is partial and case-insensitive. Everyday words work too, so "interruptible" finds spot and "sold out" finds out of stock.

Transparency and funding

Ads and sponsors: Paid placements sit at the top of the list and are always labeled as sponsored content. Sponsorship never influences the organic ranking itself.
Affiliates: Affiliate links are marked. We may earn a commission if you click them, but commissions never affect the order.
Prices: Shown in USD, converted at daily reference rates where a provider publishes in another currency. A month means 720 hours. How we estimate costs.

Vast.ai

Our sponsor

USA 33 configs 1x-4x checked <15m ago

On-Demand: from $1.73
Reserved: from $1.90
Spot: from $0.34

From $1.73 /GPU/hr On-Demand SXM

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 NVL CZ	94GB	28	189GB	Spot	$0.34	$0.34	In stock
4x H100 SXM US	320GB	256	504GB	On-Demand	$1.73	$6.94	In stock
4x H100 PCIe GB	320GB	64	756GB	Reserved (1mo)	$1.90	$7.60	In stock

In-stock and out-of-stock as of our last check at 2026-09-17T19:02Z. Not reported means we could not confirm availability.

Loading 30 more configs...

Massed Compute

In stock

USA 16 configs 1x-8x checked <15m ago

On-Demand: from $2.73
Spot: from $2.45

From $2.73 /GPU/hr On-Demand

Visit website

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 gpu_1x_h100_spot	80GB	20	128GB	Spot	$2.45	$2.45	In stock
1x H100 gpu_1x_h100	80GB	20	128GB	On-Demand	$2.73	$2.73	In stock

In-stock and out-of-stock as of our last check at 2026-09-17T19:00Z. Not reported means we could not confirm availability.

Loading 14 more configs...

Runcrate

In stock

USA 1 config 1x checked <30m ago

On-Demand: from $3.00

From $3.00 /GPU/hr On-Demand SXM

Visit website


1x H100 SXM5	80GB	20	128GB	On-Demand	$3.00	$3.00	In stock

In-stock and out-of-stock as of our last check at 2026-09-17T18:40Z. Not reported means we could not confirm availability.

Novita

In stock

USA 20 configs 1x-8x checked <15m ago

On-Demand: from $3.39
Reserved: from $2.79
Spot: from $1.70

From $3.39 /GPU/hr On-Demand SXM

Visit website

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 SXM	80GB	22	150GB	Spot	$1.70	$1.70	In stock
1x H100 SXM	80GB	22	150GB	Reserved (12mo)	$2.79	$2.79	In stock
1x H100 SXM	80GB	22	150GB	On-Demand	$3.39	$3.39	In stock

In-stock and out-of-stock as of our last check at 2026-09-17T19:00Z. Not reported means we could not confirm availability.

Loading 17 more configs...

Runpod

In stock

USA 11 configs 1x-8x checked <15m ago

On-Demand: from $3.49
Reserved: on request

From $3.49 /GPU/hr On-Demand SXM

Visit website

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 SXM Secure Cloud	80GB	26	120GB	On-Demand	$3.49	$3.49	In stock
8x H100 SXM Secure Cloud	640GB	208	960GB	Reservation	Custom	Custom	Not reported

In-stock and out-of-stock as of our last check at 2026-09-17T19:00Z. Not reported means we could not confirm availability.

Loading 9 more configs...

EmpirioLabs AI

In stock

USA 7 configs 1x-8x checked <15m ago

On-Demand: from $2.50

From $2.50 /GPU/hr On-Demand

Visit website


1x H100 h100-80gb-1x	80GB	32	126GB	On-Demand	$2.50	$2.50	In stock
1x H100 NVL h100-nvl-1x	94GB	18	128GB	On-Demand	$5.60	$5.60	In stock
2x H100 h100-80gb-2x	160GB	48	252GB	On-Demand	$2.80	$5.60	In stock
2x H100 NVL h100-nvl-2x	188GB	36	256GB	On-Demand	$5.60	$11.20	In stock
4x H100 h100-80gb-4x	320GB	96	504GB	On-Demand	$2.80	$11.20	In stock
4x H100 NVL h100-nvl-4x	376GB	72	512GB	On-Demand	$5.60	$22.40	In stock
8x H100 h100-80gb-8x	640GB	208	1.8TB	On-Demand	$4.80	$38.40	In stock

In-stock and out-of-stock as of our last check at 2026-09-17T19:00Z. Not reported means we could not confirm availability.

Lambda Labs

In stock

USA 5 configs 1x-8x checked <15m ago

On-Demand: from $4.19

From $4.19 /GPU/hr On-Demand SXM

Visit website


1x H100 PCIe gpu_1x_h100_pcie	80GB	26	200GB	On-Demand	$3.29	$3.29	Out of stock
1x H100 SXM5 gpu_1x_h100_sxm5	80GB	26	225GB	On-Demand	$4.29	$4.29	Out of stock
2x H100 SXM5 gpu_2x_h100_sxm5	160GB	52	450GB	On-Demand	$4.19	$8.38	In stock
4x H100 SXM5 gpu_4x_h100_sxm5	320GB	104	900GB	On-Demand	$4.09	$16.36	Out of stock
8x H100 SXM5 gpu_8x_h100_sxm5	640GB	208	1.8TB	On-Demand	$3.99	$31.92	Out of stock

In-stock and out-of-stock as of our last check at 2026-09-17T19:00Z. Not reported means we could not confirm availability.

Thunder Compute

In stock

USA 4 configs 1x-8x checked <15m ago

On-Demand: from $3.20

From $3.20 /GPU/hr On-Demand PCIe

Visit website


1x H100 PCIe h100_x1	80GB	4	32GB	On-Demand	$3.20	$3.20	In stock
2x H100 PCIe h100_x2	160GB	8	64GB	On-Demand	$3.20	$6.40	In stock
4x H100 PCIe h100_x4	320GB	60	480GB	On-Demand	$3.20	$12.80	Out of stock
8x H100 PCIe h100_x8	640GB	120	960GB	On-Demand	$3.20	$25.60	Out of stock

In-stock and out-of-stock as of our last check at 2026-09-17T19:00Z. Not reported means we could not confirm availability.

Koyeb

In stock

France 4 configs 1x-8x checked <30m ago

On-Demand: from $2.50

From $2.50 /GPU/hr On-Demand

Visit website


1x H100 gpu-nvidia-h100	80GB	15	180GB	On-Demand	$2.50	$2.50	In stock
2x H100 2-gpu-nvidia-h100	160GB	30	360GB	On-Demand	$2.50	$5.00	In stock
4x H100 4-gpu-nvidia-h100	320GB	60	720GB	On-Demand	$2.50	$10.00	In stock
8x H100 8-gpu-nvidia-h100	640GB	120	1.4TB	On-Demand	$2.50	$20.00	Waitlist

In-stock and out-of-stock as of our last check at 2026-09-17T18:40Z. Not reported means we could not confirm availability.

Sesterce

In stock

France 34 configs 1x-8x checked <15m ago

On-Demand: from $2.75

From $2.75 /GPU/hr On-Demand

Visit website

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100	80GB	28	180GB	On-Demand	$2.75	$2.75	In stock

Lyceum

In stock

Germany 9 configs 1x-8x checked <15m ago

On-Demand: from $2.79
Reserved: on request
Spot: from $1.10

From $2.79 /GPU/hr On-Demand

Visit website


1x H100	80GB	20	180GB	Spot	$1.10	$1.10	In stock
2x H100	160GB	40	360GB	Spot	$1.10	$2.20	Out of stock
1x H100	80GB	20	180GB	On-Demand	$2.79	$2.79	In stock
4x H100	320GB	80	720GB	Spot	$1.10	$4.40	Out of stock
2x H100	160GB	40	360GB	On-Demand	$2.79	$5.58	Out of stock
8x H100	640GB	160	1.4TB	Spot	$1.10	$8.80	Out of stock
4x H100	320GB	80	720GB	On-Demand	$2.79	$11.16	Out of stock
8x H100	640GB	160	1.4TB	On-Demand	$2.79	$22.32	Out of stock
8x H100	640GB	160	1.4TB	Reservation	Custom	Custom	Not reported

In-stock and out-of-stock as of our last check at 2026-09-17T19:00Z. Not reported means we could not confirm availability.

Scaleway

In stock

France 5 configs 1x-8x checked <30m ago

On-Demand: from $3.29

From $3.29 /GPU/hr On-Demand PCIe

Visit website


1x H100 PCIe H100-1-80G	80GB	24	240GB	On-Demand	$3.29	$3.29	In stock
2x H100 PCIe H100-2-80G	160GB	48	480GB	On-Demand	$3.29	$6.58	In stock
2x H100 SXM H100-SXM-2-80G	160GB	32	240GB	On-Demand	$3.80	$7.60	In stock
4x H100 SXM H100-SXM-4-80G	320GB	64	480GB	On-Demand	$3.67	$14.66	In stock
8x H100 SXM H100-SXM-8-80G	640GB	128	960GB	On-Demand	$3.63	$29.08	Not reported

In-stock and out-of-stock as of our last check at 2026-09-17T18:41Z. Not reported means we could not confirm availability.

Verda

In stock

Finland 35 configs 1x-8x checked <15m ago

On-Demand: from $3.30
Reserved: from $2.47
Spot: from $1.65

From $3.30 /GPU/hr On-Demand SXM

Visit website

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 SXM5 1H100.80S.32V	80GB	32	170GB	Spot	$1.65	$1.65	In stock
1x H100 SXM5 1H100.80S.30V	80GB	30	120GB	Reserved (24mo)	$2.47	$2.47	In stock
1x H100 SXM5 1H100.80S.30V	80GB	30	120GB	On-Demand	$3.30	$3.30	In stock

In-stock and out-of-stock as of our last check at 2026-09-17T19:02Z. Not reported means we could not confirm availability.

Loading 32 more configs...

GPU.ai

In stock

UAE 8 configs 1x-8x checked <15m ago

On-Demand: from $1.98

From $1.98 /GPU/hr On-Demand PCIe

Visit website

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 PCIe	80GB	--	--	On-Demand	$1.98	$1.98	In stock

Lium

In stock

Distributed 7 configs 1x-8x checked <15m ago

On-Demand: from $1.25

From $1.25 /GPU/hr On-Demand PCIe

Visit website

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 PCIe	80GB	32	126GB	On-Demand	$1.25	$1.25	In stock


1x H100 PCIe n3-H100x1-bigroot-spot__CANADA-1	80GB	28	180GB	Spot	$2.21	$2.21	In stock
1x H100 PCIe n3-H100x1-bigroot__CANADA-1	80GB	28	180GB	On-Demand	$2.65	$2.65	In stock
1x H100 SXM EU North 1::gpu-h100-sxm::1gpu-16vcpu-200gb::spot	80GB	16	200GB	Spot	$2.94	$2.94	In stock
1x H100 SXM EU North 1::gpu-h100-sxm::1gpu-16vcpu-200gb	80GB	16	200GB	On-Demand	$5.07	$5.07	In stock
8x H100 NVL n3-H100x8-NVLink-spot__CANADA-1	640GB	252	1.4TB	Spot	$2.60	$20.81	In stock

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 h100	80GB	16	80GB	Reserved (1mo)	$1.53	$1.53	In stock
1x H100 h100	80GB	16	80GB	On-Demand	$1.80	$1.80	In stock

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
8x H100 SXM5 nvidia-h100-8gpu-sxm5-cluster	640GB	96	1.5TB	Custom	Custom	Custom	In stock


8x H100 SXM5	640GB	--	2.0TB	Custom	Custom	Custom	In stock

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
8x H100 PCIe 260-H100-12mo	640GB	--	--	Reserved (12mo)	$2.09	$16.72	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 SXM	80GB	--	--	On-Demand	$1.99	$1.99	Not reported
1x H100 SXM	80GB	--	--	Reserved (6mo)	$3.19	$3.19	Not reported


1x H100 PCIe	80GB	26	200GB	On-Demand	$1.83	$1.83	Not reported
8x H100 SXM	640GB	--	--	Custom	Custom	Custom	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 HOPPER_H100	80GB	--	--	On-Demand	$3.40	$3.40	Not reported


1x H100 SXM5	80GB	16	32GB	On-Demand	$2.36	$2.36	Not reported


1x H100 PCIe	80GB	--	--	On-Demand	$4.00	$4.00	Not reported


1x H100	80GB	--	--	Custom	$1.89	$1.89	Not reported
1x H100	80GB	--	--	On-Demand	$4.50	$4.50	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 SXM	80GB	--	--	Reserved (12mo)	$3.26	$3.26	Not reported
1x H100 SXM gpu-h100x1-80gb	80GB	20	240GB	On-Demand	$4.41	$4.41	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 gpu-h100	80GB	13	144GB	On-Demand	$5.49	$5.49	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 NVL NCadsH100v5	94GB	40	320GB	Spot	$1.29	$1.29	Not reported
1x H100 NVL NCadsH100v5	94GB	40	320GB	Reserved (36mo)	$3.84	$3.84	Not reported
1x H100 NVL NCadsH100v5	94GB	40	320GB	On-Demand	$6.98	$6.98	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 SXM H100	80GB	20	250GB	On-Demand	$5.95	$5.95	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 p5.4xlarge	80GB	16	256GB	On-Demand	$6.88	$6.88	Not reported
8x H100 p5.48xlarge	640GB	192	2.0TB	Reserved (36mo)	$2.97	$23.78	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
2x H100 SXM	160GB	32	464GB	On-Demand	$1.75	$3.51	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100	80GB	--	--	Reservation	$1.75	$1.75	Not reported
1x H100 PCIe	80GB	--	--	Spot	$2.00	$2.00	Not reported
1x H100	80GB	28	180GB	On-Demand	$2.50	$2.50	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 h100-380	80GB	30	380GB	On-Demand	$2.99	$2.99	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 SXM	80GB	23	184GB	Reserved (36mo)	$2.49	$2.49	Not reported
1x H100 SXM	80GB	23	184GB	On-Demand	$2.99	$2.99	Not reported


1x H100 SXM NVIDIA HGX H100	80GB	16	200GB	Spot	$2.15	$2.15	Not reported
1x H100 SXM NVIDIA HGX H100	80GB	16	200GB	On-Demand	$3.85	$3.85	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 SXM a3-highgpu-1g	80GB	26	234GB	Reserved (36mo)	$4.86	$4.86	Not reported
2x H100 SXM a3-highgpu-2g	160GB	52	468GB	On-Demand	$11.07	$22.14	Not reported
8x H100 SXM a3-megagpu-8g	640GB	208	1.9TB	Spot	$3.77	$30.15	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
2x H100 SXM5	160GB	40	400GB	On-Demand	$3.25	$6.50	Not reported


8x H100 SXM	640GB	128	2.0TB	Spot	$2.46	$19.71	Not reported
8x H100 SXM	640GB	128	2.0TB	On-Demand	$6.16	$49.24	Not reported


8x H100 BM.GPU.H100.8	640GB	--	--	On-Demand	$10.00	$80.00	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
8x H100 SXM 8H100.128v.2048m SXM-12mo	640GB	128	1.5TB	Reserved (12mo)	$2.15	$17.16	Not reported
8x H100 SXM 8H100.128v.2048m SXM	640GB	128	1.5TB	On-Demand	$3.33	$26.60	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100	80GB	36	256GB	Reserved (24mo)	$2.92	$2.92	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 PCIe H100_PCIE_80GB_x1	80GB	28	180GB	Reserved (36mo)	$2.56	$2.56	Not reported
1x H100 PCIe H100_PCIE_80GB_x1	80GB	28	180GB	On-Demand	$3.50	$3.50	Out of stock

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
8x H100	640GB	96	2TB	Reserved (12mo)	$3.47	$27.78	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
3x H100 PREMIUM_DL385G11-9334-3xH100	240GB	64	128GB	Reserved (1mo)	$1.92	$5.75	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 SXM N.HGXH100.250	80GB	26	250GB	Reserved (12mo)	$3.31	$3.31	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 SXM	80GB	--	--	Custom	Custom	Custom	Not reported


8x H100 SXM h100-80gb-sxm-ib.8x	640GB	176	960GB	On-Demand	$3.90	$31.20	Out of stock
8x H100 SXM h100-80gb-sxm-ib.8x	640GB	176	960GB	Spot	Custom	Custom	Not reported
8x H100 SXM h100-80gb-sxm-ib.8x	640GB	176	960GB	Reservation	Custom	Custom	Not reported


8x H100 SXM	640GB	--	--	Custom	Custom	Custom	Not reported

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 SXM gpu_1x_h100	80GB	20	128GB	On-Demand	$3.00	$3.00	Out of stock

Heads up: A provider's own page may quote a different figure, for example on tax, region or a promotion. A monthly-only plan shows a derived hourly rate. Verify before provisioning. More on how we price.

What can the H100 run?

One H100 has 80 GB of VRAM on the SXM variant. In practice, that's enough memory for roughly 120B parameters at 4-bit or 31B at 16-bit, assuming a 32K context. Below are some open-weight LLMs, with the estimated memory and GPUs each one needs.

Fits on 1 Single node ≤ 8 Multi-node

Model	Memory (INT4 / FP4)(FP8 / INT8)(FP16 / BF16)	H100s needed	Cost /hr	Cost /mo
Qwen3.8-27B 27B	18 GB30 GB58 GB	1	$3.38$3.38$3.38	$2,434$2,434$2,434
Gemma 4 31B 30.7B	22 GB35 GB69 GB	1	$3.38$3.38$3.38	$2,434$2,434$2,434
GPT-OSS-120B 117B · 5.1B active	67 GB120 GB237 GB	1	$3.38$6.76$13.52	$2,434$4,867$9,734
GLM-5.3-Flash 320B · 18B active	178 GB322 GB642 GB	4	$13.52$27.04–	$9,734$19,469–
MiniMax-M3 428B · 23B active	239 GB432 GB862 GB	4	$13.52$27.04–	$9,734$19,469–
DeepSeek V4.1 Flash 552B · 16B active	306 GB554 GB1,106 GB	8	$27.04$27.04–	$19,469$19,469–
Kimi K3 2.8T · 104B active	1,544 GB2,804 GB5,605 GB	22	–––	–––

Estimates based on the median on-demand rate. Memory is weights plus FP8 KV cache at 32K context per request (FP16/BF16 in the 16-bit column). GPU counts assume 90% of advertised VRAM is usable. No guarantee of runtime support, usable performance, or that a matching quantized build exists. How we estimate costs.

Vast.ai

Per-second billing. Pay only for what you use.

50+ GPU models from RTX 3060 to B200, from $0.04/hr

No minimums, no long-term contracts.

SOC2 certified · PyTorch, CUDA & TensorFlow templates ready

Learn more

Technical Specifications

Nvidia H100 · Per GPU · SXM

Compute · dense
FP8	1,979 TFLOPS 3,958 with sparsity
FP16 / BF16	989.5 TFLOPS 1,979 with sparsity
INT8	1,979 TOPS 3,958 with sparsity
FP32	67 TFLOPS
FP64	34 TFLOPS
Precision support	FP8FP16BF16TF32FP32FP64INT8

Memory
Capacity	80 GB HBM3
Bandwidth	3,350 GB/s
Bus width	5,120-bit
ECC	Yes

Silicon
Architecture	Hopper
Process	TSMC 4N
Transistors	80 billion
Shader cores	16,896 CUDA cores
Matrix cores	528 Tensor cores
Compute units	132 SMs

Fabric and host
GPU interconnect	NVLink 900 GB/s
Host interface	PCIe 5.0 x16

Power
Board power	700 W

Platform
Partitioning	MIG, up to 7 instances

Source: official Nvidia H100 datasheet.

Variants

The H100 ships in 3 versions with different memory, bandwidth or interconnect. The figures at the top of this page are for the H100 SXM.

Variant	Memory	Memory bandwidth	Interconnect
H100 SXM	80 GB	3,350 GB/s	NVLink 900 GB/s
H100 PCIe	80 GB	2,000 GB/s	NVLink 600 GB/s
H100 NVL	94 GB	3,938 GB/s	NVLink 600 GB/s

Frequently Asked Questions

How much does the H100 cost per hour?

As of September 17, 2026, the median on-demand price is $3.38 per GPU per hour across 38 providers with a priced on-demand config.

Billing type	Configs	Median /GPU/hr	Cheapest
On-demand	201	$3.38	$1.25 (in stock, Lium)
Reserved	176	$3.13	$1.53 (1 mo, in stock, HyperAI)
Spot	49	$2.00	$0.34 (in stock, Vast.ai)

6 providers also quote the H100 on a custom contract, priced per deal.

Who has the cheapest H100?

The cheapest verified in-stock on-demand H100 tracked by GetDeploying is $1.25 per GPU per hour from Lium. Cheapest verified in stock on other billing types: reserved $1.53 (1 mo) from HyperAI and spot $0.34 from Vast.ai.

How much does the H100 cost per month?

At 720 hours per month, one H100 costs an estimated $2,434 at the median on-demand price. Cheapest verified in stock: $900 per month on-demand, $1,102 reserved (1 mo), $245 spot.

Where can I rent an H100?

GetDeploying currently tracks H100 configs from 53 providers. The cheapest verified in-stock on-demand configs come from Lium, Vast.ai, HyperAI and GPU.ai. See the full price comparison above for every provider and config.

Are H100 prices going up or down?

As of September 17, 2026, the median on-demand price has risen about 7% over the past 90 days to $3.38 per GPU per hour, and about 11% over the past 12 months.

How many H100 GPUs do I need?

One H100 runs models up to roughly 120B parameters at 4-bit quantization or 31B at 16-bit, assuming a 32K context. Larger models run across multiple GPUs: the model table above shows the estimated memory and GPU count for popular open-weight LLMs.

Why choose the H100?

80GB HBM3 with 900 GB/s NVLink for multi-node training. FP8 Transformer Engine for improved training throughput over A100. The standard choice for large-scale AI training and high-throughput inference.

When is the H100 not a good fit?

May be more than needed for single-GPU inference on models under 30B parameters. Consider the L40S or A10 for cost-effective inference workloads.

Alternatives to Nvidia H100

Nvidia B300 288 GB HBM3e · 2.4x the bandwidth of the H100
From $4.95 /GPU/hr Compare
Nvidia H200 141 GB HBM3e · 43% more bandwidth than the H100
From $2.09 /GPU/hr Compare
Nvidia RTX 5090 32 GB GDDR7 · 47% less bandwidth than the H100
From $0.35 /GPU/hr Compare
Nvidia A100 80 GB HBM2e · 42% less bandwidth than the H100
From $0.40 /GPU/hr Compare
AMD MI300X 192 GB HBM3 · 58% more bandwidth than the H100
From $2.59 /GPU/hr Compare

Browse all GPUs

Last updated Sept. 17, 2026

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 GPU-SPOT-12xCPU-240GB-1xH100	80GB	12	240GB	Spot	$1.88	$1.88	Out of stock
1x H100 GPU-12xCPU-240GB-1xH100	80GB	12	240GB	On-Demand	$1.89	$1.89	Out of stock

Configuration	VRAM	vCPU	RAM	Billing	Per GPU/h	Total/h	Availability
1x H100 g3-h100-small-12mo	80GB	16	192GB	Reserved (12mo)	$1.18	$1.18	Out of stock
1x H100 g3-h100-small	80GB	16	192GB	On-Demand	$3.37	$3.37	Out of stock