H100 cloud pricing per hour and rental comparison
As of , the cheapest single H100 confirmed in stock is $1.10 per GPU-hour on Lium. The lowest listed price is $1.00 per GPU-hour on Lium, but that provider does not expose a current stock signal. GPU Finder compares 18 providers for this size and refreshes availability hourly. Methodology.
Compare H100 on-demand and spot rental rates across 23 cloud providers, with 8x H100 node pricing and live stock.
Structured offer basis: 89 current 1× H100 listings range from $1.00 to $15.18 USD per physical GPU-hour. Whole-node prices for other GPU counts are not mixed into this range.
Cheapest single H100 rental
verifiedCheapest H100 confirmed in stock right now
Limited- Is now a good time to rent an H100 cloud GPU?
- ■ mid-range vs the last 6 months — The 8× H100 on-demand floor is $12.40/hr ($1.55/GPU-hr), down 3% vs last month — 6-month range $2.33–$15.20/hr for 8× nodes. In short, prices are mid-range versus the last 6 months.
- What is the cheapest H100 cloud GPU available right now?
- Cheapest H100 confirmed in stock right now: $1.10/GPU-hr on Lium (1× listing, $1.10/hr).
- Which cloud providers most reliably have the H100 in stock?
- Most reliably in stock over the last 30 days: PrimeIntellect, Shadeform, Digital Ocean. Frequently waitlisted: Vast, Cudo.
Prices are the cheapest listed on-demand rates; “in stock” means the provider reported capacity on our last hourly check. How we compute this →
H100 rental prices by GPU count
Lowest on-demand H100 price reported in stock at each node size across 23 providers, normalised to per-GPU-hour so 1× and 8× listings compare like for like. Refreshed hourly.
| Node | Per GPU-hour | Node $/hr | Provider | Region | Stock | Spot / GPU-hr |
|---|---|---|---|---|---|---|
| 1× H100 | $1.10 | $1.10 | Lium18 providers at 1× | The Netherlands | Limited | $0.34 |
| 2× H100 | $1.55 | $3.10 | Lium15 providers at 2× | Russia | Limited | $0.01 |
| 4× H100 | $1.55 | $6.20 | Lium12 providers at 4× | Russia | Limited | $0.01 |
| 8× H100 | $2.53 | $20.27 | Vast16 providers at 8× | United Arab Emirates, AE | Limited | $0.26 |
H100 in stock now
- Lium $1.10/GPU-hr · 1× · 21% 30d
- Vast $1.73/GPU-hr · 2× · 0% 30d
- Cudo $1.82/GPU-hr · 1× · 0% 30d
- PrimeIntellect $1.90/GPU-hr · 1× · 100% 30d
- Scaleway $2.52/GPU-hr · 1× · 63% 30d
- Runpod $2.69/GPU-hr · 1× · 42% 30d
Cheapest in-stock configuration per provider; “in stock” means the provider reported capacity on our last hourly check. Full list in the pricing table and the availability heatmap.
When to upgrade from the H100 to the H200
The H200 is the same Hopper GH100 die as the H100 — identical 1,979 TFLOPS FP16, NVLink 4.0 and 700 W — with the memory swapped from 80 GB HBM3 to 141 GB HBM3e (76% more capacity, 43% more bandwidth). Upgrade when memory is the constraint: serving a 70B-class model in FP16 that spills to two H100s but fits one H200, long-context inference where the KV cache dominates, or batch sizes capped by VRAM. In those cases one H200 often replaces two H100s and the per-token cost drops even though the hourly rate is 60–70% higher.
Stay on the H100 when the job is compute-bound — pre-training, fine-tuning of models that already fit, and anything that saturates the Tensor Cores at 80 GB — because the H200 runs the same math at the same speed for more money. H100 spot is also far deeper than H200 spot, so price-sensitive, interruptible training is still an H100 workload. The live H100 vs H200 comparison page has today's prices and stock side by side; the blog guide walks through the scenario maths.
H100 rental prices by provider
- hyperstack__1xH1…80GB__28__180In stock100% · 30d$1.90/hr1× H10028 vCPU180 GBFree egressCA
Get notified when H100 drops in price or reliability changes
One email per change, max once a day. We send a confirmation link first; one-click unsubscribe in every email.
H100 (1× GPU) availability — last 7 days
3 of 12 providers kept 1× H100 ≥80% available all week. Polled hourly · 12 providers tracked · 7-day window.
| Provider | Fri11 | Sat12 | Sun13 | Mon14 | Tue15 | Wed16 | Thu17 | Week |
|---|---|---|---|---|---|---|---|---|
| Digital Ocean | 100% | |||||||
| Nebius | 100% | |||||||
| PrimeIntellect | 100% | |||||||
| Scaleway | 68% | |||||||
| Shadeform | 58% | |||||||
| Lambda | 24% | |||||||
| Hyperstack | 18% | |||||||
| Runpod | 13% | |||||||
| Lium | 4% | |||||||
| AceCloud | 0% | |||||||
| Cudo | 0% | |||||||
| Vast | 0% |
30-day reliability depth
GPU Finder does not treat availability as a static yes/no flag. For H100, current stock badges are paired with a 30-day reliability score based on our hourly stock checks — the share of tracked time each listing was reported available. How we compute this →
- 170
- listings with a 30-day score
- 58%
- of listings on this page have a score
- 30d
- longest stock history on this page
Strongest reliability signals here: PrimeIntellect (100%, 30d covered), Shadeform (100%, 30d covered), Digital Ocean (100%, 30d covered).
Caveat: scores stay hidden until a listing has at least 48 tracked hours, and some provider APIs expose coarse capacity levels instead of exact stock counts. Use the score as historical depth beside current availability, not as a guarantee that a GPU will still be allocatable when you click through.
Price History & Comparison
Full H100 price history (20 months) →Spot pricing decision guide
Treat spot as a risk-adjusted capacity decision, not just a cheaper number.
per GPU-hour
99% below on-demand floor
daily median for the same offers with complete history
median $2.60/GPU-hr
Cheapest trusted spot / interruptible capacity is on Vast; 2 providers currently report spot or interruptible stock (41 units/signals visible).
Nebius has kept spot or interruptible capacity in stock most reliably over the last 30 days (99%). See the availability section for the broader stock history.
The discount is visible, but availability, reliability, or volatility argues for more caution.
Good fit: checkpointed batch jobs, flexible training, and experiments that can restart. Avoid for production serving, deadline-bound runs, or jobs that cannot tolerate eviction.
Source caveat: Runpod exposes explicit spot fields; Vast is labeled interruptible/bid-floor. Lambda is treated as on-demand-only unless a verified spot field is added.
H100 spot and interruptible pricing
H100 spot and interruptible GPU pricing can vary by provider, stock, and interruption risk. GPU Finder shows verified Runpod spot rows and Vast interruptible/bid-related rows beside on-demand pricing where current source data supports it.
Vast interruptible
$0.01/GPU-hr
vs on-demand $3.27/GPU-hr · 100% below on-demand
Vast interruptible H100 rowsGood fit: checkpointed batch jobs, flexible training, and experiments that can restart. Avoid for production serving, deadline-bound runs, or jobs that cannot tolerate eviction. Compare Runpod vs Vast spot and interruptible before renting.
H100 spot and interruptible FAQ
Who has H100 spot GPU pricing?
GPU Finder shows verified Vast interruptible/bid-related H100 rows where current source data supports it. Check the pricing table for on-demand and spot or interruptible offers from other providers.
What is the difference between H100 spot and interruptible pricing?
Runpod exposes explicit spot fields tied to Secure Cloud and Community Cloud. Vast uses a marketplace model, so its discounted H100 capacity is labeled interruptible or bid-related rather than generic spot.
Should I choose the cheapest H100 interruptible offer?
Only if the workload can tolerate interruption. H100 spot and interruptible prices are best for checkpointed batch jobs, flexible training, and experiments. Avoid them for production serving, deadline-bound runs, or non-checkpointed workloads.
About the H100
The NVIDIA H100 is the GPU that defined the generative AI era. Built on the Hopper architecture with a dedicated Transformer Engine, it accelerates large language model training and inference at a scale no previous GPU could match. It remains the default choice for teams training frontier models and any workload where FP8 throughput and NVLink bandwidth are bottlenecks.
Key Specifications
| Architecture | Hopper (GH100) |
| GPU Memory | 80 GB HBM3 (SXM5) |
| Memory Bandwidth | 3.35 TB/s |
| FP16 Tensor Core | 1,979 TFLOPS |
| TDP | 700W (SXM5) / 350W (PCIe) |
| Interconnect | NVLink 4.0 (900 GB/s) |
| Release Year | 2023 |
Cloud Pricing Context
H100 on-demand pricing ranges from roughly $1.35/hr on budget neoclouds to $8+/hr on hyperscalers. The market median sits around $2.45/hr across 50+ configurations. Spot instances are widely available at 30-50% discounts. Prices have declined steadily since mid-2024 as B200 supply ramps up.
Best For
- Pre-training and fine-tuning LLMs at the 7B-70B parameter scale
- High-throughput inference serving with FP8 quantization via the Transformer Engine
- Multi-node distributed training using NVLink and NVSwitch fabrics
- Production inference pipelines requiring predictable latency at scale
H100 by node size
Dated availability research
Cite the GPU Cloud Availability Index for frozen H100 provider reliability rankings, coverage hours and downloadable data.