Try
10 GPU-hours
$9.20
$0.92/hour
- Test your setup end to end
- Same GPU and image as a long run
- Unused hours never expire
Quick answer
How much does an A100 cost per hour?
As of 29 October 2026, an NVIDIA A100 80GB costs from $1.09 to $5.07 per GPU-hour on demand across 13 providers, with a median of $2.23. The A100 40GB starts at $0.89, with a median of $1.99. On GPUwithCrypto an A100 80GB costs $0.92/hour, billed per minute and paid in crypto with no KYC.
How much does an A100 80GB cost per hour at each provider?
One row per provider at its cheapest on-demand A100 80GB configuration, sorted by price. The most expensive on-demand A100 80GB we found (Google Cloud, $5.07) costs 4.7 times the cheapest ($1.09 at Thunder Compute).
| Provider | Configuration | On-demand $/GPU-hour | Spot / marketplace $/GPU-hour | Billing increment | Accepts crypto | ID / KYC |
|---|---|---|---|---|---|---|
| GPUwithCrypto | A100 80GB, 1 to 8 GPUs; SXM or PCIe shown per host | $0.92 | — (on-demand only) | Per minute | Yes | No |
| Thunder Compute | A100 80GB (form factor not stated), 1 GPU | $1.09 | — | Per minute | No | None documented |
| Massed Compute | A100 80GB (form factor not stated), 1 GPU (SXM: $1.38) | $1.35 | — | Hourly rate (increment not stated) | Yes (BTC) | None documented |
| Jarvislabs | A100 80GB (form factor not stated), 1 GPU | $1.49 | $0.89 (spot) | Per minute | No | None documented |
| Spheron | A100 80GB (form factor not stated) | $1.69 | — | Per minute, 20-minute minimum | Yes (USDC) | None documented |
| RunPod | A100 80GB PCIe or SXM, 1 GPU | $1.79 | $1.19 PCIe, $1.39 SXM (Community Cloud) | Per second | Yes | KYC before first crypto payment |
| Crusoe | A100 80GB PCIe (SXM: $2.30) | $2.00 | Contact sales | Per hour or per second | No | None documented |
| Verda (formerly DataCrunch) | A100 80GB SXM, 1 GPU | $2.23 | $1.11 (spot) | — | No | None documented |
| Vultr | A100 80GB (form factor not stated), 1 GPU | $2.40 | — | Hourly | Yes (BitPay) | None documented |
| CoreWeave | A100 80GB, 8-GPU instance ÷ 8 | $2.70 | $1.21 (spot) | — | Unknown | Unknown |
| Lambda | A100 80GB SXM, 8 GPUs | $2.79 | — | Per minute | No (cards only) | None documented |
| OVHcloud | A100 80GB (form factor not stated), 1 GPU | $3.07 | — | Per second, 1-minute minimum | No | Sometimes |
| AWS | A100 80GB, 8 GPUs (p4de.24xlarge) ÷ 8 | $3.43 | — | — | No | None documented |
| Google Cloud | A100 80GB, 1 GPU (a2-ultragpu-1g) | $5.07 | $3.01 (spot) | Per second, 1-minute minimum | No | None documented |
| Vast.ai | A100 80GB PCIe and SXM, marketplace | — | $0.34 PCIe, $0.48 SXM (Vast.ai's public price feed, cheapest rentable offers) | Per second | Yes (BitPay, Crypto.com) | None documented |
| Akash Network | A100 80GB SXM, decentralized marketplace | — | $1.07 (lowest recent provider bid) | — | Yes (AKT, ACT) | None documented |
| Shadeform | A100 80GB SXM, 1 GPU, resold from other clouds | — | $1.50 (cheapest available listing) | — | Yes (USDC, per Stripe) | None documented |
Prices checked on
USD per GPU-hour, taxes excluded, rounded to the cent. Hyperscaler prices cover the whole VM (AWS US East, Google Cloud Iowa). Vultr's figure comes from its public plans API, which did not confirm availability. Paperspace showed only a 3-year commitment price for its A100 80GB, so it is not listed. 'None documented' means no ID step in the official pages we read.
A100 40GB vs 80GB: how much more does the 80GB cost?
According to GPUwithCrypto's price check of 29 October 2026, the same provider charges 25% to 67% more for the 80GB A100 than the 40GB. The medians differ by only 12%, because the cheapest clouds list only the 80 GB card.
| Provider | A100 40GB | A100 80GB | 80GB premium |
|---|---|---|---|
| Jarvislabs | $0.89 | $1.49 | +67% |
| Verda (formerly DataCrunch) | $1.59 (SXM) | $2.23 (SXM) | +40% |
| Lambda | $1.99 (SXM, 8 GPUs) | $2.79 (SXM, 8 GPUs) | +40% |
| Google Cloud | $3.67 | $5.07 | +38% |
| AWS | $2.74 (p4d, 8 GPUs ÷ 8) | $3.43 (p4de, 8 GPUs ÷ 8) | +25% |
| Median of all providers | $1.99 (5 providers) | $2.23 (13 providers) | +12% |
Prices checked on
The 80 GB card also has faster memory: 2,039 GB/s on the SXM version against 1,555 GB/s for the 40 GB SXM4. Compute is the same.
Where is an A100 cheaper than ours, and why?
On marketplaces, on spot, and in the 40 GB version. We say so because the cheapest option depends on your job.
- Marketplace: the cheapest rentable A100 80GB PCIe in Vast.ai's public price feed cost $0.34 per hour, among 91 rentable offers. Independent hosts set these prices and run the machines.
- Spot: Jarvislabs sells interruptible A100 80GB hours at $0.89. The provider can stop them.
- 40 GB: Jarvislabs' on-demand A100 40GB costs $0.89. If your model fits in 40 GB, it is cheaper than our 80 GB card.
For an 80 GB A100 that nobody can take back, our $0.92/hour is 59% below the median of 13 providers. Instances are on-demand, not spot: nobody can take your GPU back while your balance covers it, and the instance runs until you stop it. The reference is Thunder Compute's A100 80GB at $1.09, whose page does not state the form factor.
How much does an A100 cost per day, week or month?
Our A100 80GB price next to the lowest and the median self-serve on-demand price of 29 October 2026, for the same hours.
| Period | GPU-hours | GPUwithCrypto | Lowest elsewhere (Thunder Compute) | Median elsewhere |
|---|---|---|---|---|
| 1 hour | 1 | $0.92 | $1.09 | $2.23 |
| 1 day | 24 | $22.08 | $26.16 | $53.42 |
| 1 week | 168 | $154.56 | $183.12 | $373.97 |
| 30 days | 720 | $662.40 | $784.80 | $1,602.72 |
Prices checked on
The median is $2.226 per GPU-hour before rounding. Our day, week and month match the 24, 168 and 720 GPU-hour blocks.
What are the A100's specs?
NVIDIA's official figures. NVIDIA prints both dense and sparse Tensor values for the A100; compare the dense ones with other GPUs.
| Spec | A100 80GB SXM | A100 80GB PCIe | A100 40GB SXM4 |
|---|---|---|---|
| Memory | 80 GB HBM2e | 80 GB HBM2e | 40 GB HBM2 |
| Memory bandwidth | 2,039 GB/s | 1,935 GB/s | 1,555 GB/s |
| FP16 / BF16 Tensor, dense | 312 TFLOPS (624 sparse) | 312 TFLOPS (624 sparse) | 312 TFLOPS (624 sparse) |
| TF32 Tensor, dense | 156 TFLOPS (312 sparse) | 156 TFLOPS (312 sparse) | 156 TFLOPS (312 sparse) |
| FP8 Tensor | Not supported | Not supported | Not supported |
| GPU-to-GPU link | NVLink 600 GB/s | NVLink bridge 600 GB/s | NVLink 600 GB/s |
| Max power | 400 W | 300 W | 400 W |
What fits on one A100 40GB or 80GB?
Memory needed for common open models, using weights × 1.2 for inference (EleutherAI's rule of thumb). The A100 has no FP8 Tensor Cores; the L40S, H100 and H200 do. The VRAM guide has the formulas.
| Model and job | Memory needed | A100 40GB | A100 80GB |
|---|---|---|---|
| Llama 3.1 8B inference, BF16 | ≈ 19.2 GB | ✓ | ✓ |
| Llama 3.1 8B LoRA fine-tune | ≈ 19.2 GB | ✓ | ✓ |
| Mistral 7B inference, BF16 | ≈ 17.5 GB | ✓ | ✓ |
| FLUX.1 [dev] transformer, BF16 weights | 24 GB, plus text encoders and VAE | ✓ | ✓ |
| Llama 3.1 70B inference, 4-bit | ≈ 42 GB | ✗ | ✓ |
| Qwen2.5 72B inference, 4-bit | ≈ 43.6 GB | ✗ | ✓ |
| Llama 3.1 70B QLoRA fine-tune | ≈ 52.5 GB | ✗ | ✓ |
| Llama 3.1 8B full fine-tune | ≈ 128 to 144 GB before activations | ✗ | ✗ |
| Llama 3.1 70B inference, BF16 | ≈ 168 GB | ✗ | ✗ |
What does an A100 80GB cost on GPUwithCrypto?
$0.92/hour, at least 15% under its named on-demand reference: Thunder Compute at $1.09, read on 29 October 2026. We do not claim to be the cheapest for every GPU; spot and marketplace offers (Vast.ai) can be lower. Instances are billed per minute at the hourly rate while they run, from the GPU-hours you bought for that GPU. Stopping an instance stops the charge at once. GPU-hours are sold in blocks of 10, 24, 168 or 720 hours for one GPU model, at the same hourly rate; instances of that GPU draw from them per minute, and unused hours never expire.
Popular
Day
24 GPU-hours
$22.08
$0.92/hour
- One day of continuous running
- A fine-tuning run or a batch job
- Unused hours never expire
Week
168 GPU-hours
$154.56
$0.92/hour
- One week of continuous running
- Training runs and evaluations
- Unused hours never expire
Month
720 GPU-hours
$662.40
$0.92/hour
- 30 days of continuous running
- An inference server running all month
- Unused hours never expire
In every plan: Whole GPUs, never shared · Root SSH, CUDA and PyTorch ready · Billed per minute · Unused hours never expire
A100 price questions
What is the cheapest A100 rental?
On 29 October 2026 the lowest self-serve on-demand A100 80GB price we found was $1.09 at Thunder Compute, and GPUwithCrypto charges $0.92. Jarvislabs' A100 40GB cost $0.89. Marketplace offers in Vast.ai's public price feed started at $0.34, but hosts set and can change those prices.
A100 40GB vs 80GB: which should I rent?
Rent the 80 GB card for 70B models in 4-bit or QLoRA, which need 42 to 53 GB. The 40 GB card is enough for 7B and 8B models in BF16. The same provider charged 25% to 67% more for the 80 GB version on 29 October 2026.
How much does an A100 cost per month?
720 GPU-hours of an A100 80GB cost $662.40 on GPUwithCrypto. At the median on-demand price of 29 October 2026 the same month cost $1,602.72, and $784.80 at the lowest.
Is the A100 still worth renting?
Yes for BF16 training and inference that fits in 80 GB: it was the cheapest 80 GB data-center GPU in our 29 October 2026 check. For FP8 models pick an L40S, H100 or H200. See L40S vs A100 and H100 vs A100.
Can I rent an A100 with crypto?
Yes. GPUwithCrypto takes BTC, USDT, USDC, ETH, SOL, LTC with no ID, no selfie and no card. Of the 13 self-serve A100 80GB providers we checked, 4 accept crypto: Massed Compute, Spheron, Vultr and RunPod. RunPod asks for KYC before the first crypto payment.
Sources (20)
- Thunder Compute — Pricing thundercompute.com/pricing — Thunder Compute (read 2026-10-29)
- Massed Compute — VM pricing vm.massedcompute.com/pricing — Massed Compute (read 2026-10-29)
- Jarvislabs — Pricing jarvislabs.ai/pricing — Jarvislabs (read 2026-10-29)
- Spheron — Pricing spheron.network/pricing — Spheron (read 2026-10-29)
- Runpod — Pricing runpod.io/pricing — RunPod (read 2026-10-29)
- Crusoe — Cloud pricing crusoe.ai/cloud/pricing — Crusoe (read 2026-10-29)
- Verda — Pricing verda.com/pricing — Verda (read 2026-10-29)
- Vultr — Public plans API api.vultr.com/v2/plans?type=all&per_page=500 — Vultr (read 2026-10-29)
- CoreWeave — Pricing coreweave.com/pricing — CoreWeave (read 2026-10-29)
- Lambda — Pricing lambda.ai/pricing — Lambda (read 2026-10-29)
- OVHcloud — Public Cloud prices ovhcloud.com/en/public-cloud/prices — OVHcloud (read 2026-10-29)
- AWS — EC2 on-demand price list, US East (N. Virginia) — Amazon Web Services (read 2026-10-29)
- Google Cloud — Accelerator-optimized VM pricing — Google Cloud (read 2026-10-29)
- Vast.ai public GPU pricing feed — Vast.ai (read 2026-10-29)
- Akash Network — GPU prices API console-api.akash.network/v1/gpu-prices — Akash Network (read 2026-10-29)
- Shadeform — Instance types API api.shadeform.ai/v1/instances/types — Shadeform (read 2026-10-29)
- Runpod — Billing information docs.runpod.io/references/billing-information — RunPod (read 2026-10-29)
- NVIDIA A100 Tensor Core GPU — NVIDIA (read 2026-10-29)
- NVIDIA Ampere architecture whitepaper (A100 40GB) — NVIDIA (read 2026-10-29)
- Transformer Math 101 — EleutherAI (read 2026-10-29)