Vast.ai vs Runpod: which GPU cloud is cheaper?
We earn commissions when you shop through the links below.
Short answer (checked Nov 5, 2026 ~10:06 PM PT): the cheapest Vast.ai host beat Runpod Community Cloud on 7 of 8 overlapping cards we could sample (RTX 3090, 4090, 5090, L40S, A100 PCIe, A100 SXM, H100 SXM). Runpod Community stayed cheaper than the Vast lowest only on the RTX A6000 ($0.33 vs $0.43). The typical (median) Vast.ai offer beat Runpod Community only on the two A100 rows; for the RTX 3090, 4090, 5090, A6000 and H100 SXM, Runpod Community's fixed list price was lower than the Vast median. So: shop the Vast marketplace if you want the absolute lowest host; pick Runpod if you want one published price without comparing hosts.
This page is for people searching vast.ai vs runpod or runpod vs vast. Prices are USD per GPU per hour, on-demand only. Full provider table: GPU rental prices. Cheapest card by VRAM tier: cheapest GPU rental. Same-GPU tool view: GPU cost per hour.
Side-by-side hourly prices (Nov 5, 2026)
| GPU | VRAM | Runpod Community | Runpod Secure | Vast.ai lowest | Vast.ai median | Vast offers read |
|---|---|---|---|---|---|---|
| RTX 3090 | 24 GB | $0.22 | $0.50 | $0.18 | $0.27 | 35 |
| RTX 4090 | 24 GB | $0.34 | $0.89 | $0.30 | $0.46 | 39 |
| RTX 5090 | 32 GB | $0.69 | $1.19 | $0.43 | $0.77 | 54 |
| RTX A6000 | 48 GB | $0.33 | $0.59 | $0.43 | $0.47 | 4 |
| L40S | 48 GB | $0.79 | $1.09 | $0.60 | $0.80 | 7 |
| A100 PCIe | 80 GB | $1.19 | $1.79 | $0.87 | $1.00 | 3 |
| A100 SXM | 80 GB | $1.39 | $1.79 | $1.08 | $1.19 | 6 |
| H100 PCIe | 80 GB | $1.99 | $2.89 | — | — | 0 |
| H100 SXM | 80 GB | $2.69 | $3.99 | $2.20 | $2.80 | 6 |
H100 PCIe: Vast.ai had no matching verified on-demand single-GPU offers in this sample, so we do not invent a Vast figure. Runpod still lists both Community and Secure rates.
What a 10-hour job costs (GPU time only)
| GPU | Runpod Community | Runpod Secure | Vast.ai lowest | Vast.ai median |
|---|---|---|---|---|
| RTX 3090 | $2.20 | $5.00 | $1.80 | $2.70 |
| RTX 4090 | $3.40 | $8.90 | $3.00 | $4.60 |
| RTX 5090 | $6.90 | $11.90 | $4.30 | $7.70 |
| RTX A6000 | $3.30 | $5.90 | $4.30 | $4.70 |
| A100 SXM 80 GB | $13.90 | $17.90 | $10.80 | $11.90 |
| H100 SXM 80 GB | $26.90 | $39.90 | $22.00 | $28.00 |
Storage and data transfer are extra on both. Confirm the live rate on the provider before you rent — Vast hosts change by the minute, and Runpod stock varies by region.
Marketplace vs Secure Cloud vs Community Cloud
| Vast.ai | Runpod Secure Cloud | Runpod Community Cloud | |
|---|---|---|---|
| What it is | Peer marketplace: each host sets price, disk and bandwidth | Runpod data-center pods with a fixed price list | Runpod pods on third-party hosts; same UI, lower list price |
| Price style | Lowest + median of live offers (we sample up to 64 verified on-demand single-GPU offers per card) | One Secure rate per GPU on runpod.io/pricing | One Community rate per GPU on the same page |
| Reliability trade-off | Host quality varies; check reviews, reliability and bandwidth $/GB | Aimed at steadier data-center capacity | Cheaper, shared / community hosts |
| When it wins tonight | Lowest host on most consumer + A100/H100 SXM rows | When you want Secure and will pay the Secure premium | Predictable price that often beats the Vast median |
Vast also sells interruptible instances (often much cheaper, can be paused). We do not mix interruptible into the on-demand table above.
Every number on this page links to its source.
How we calculate + sources ⓘ
Sources and checked dates
- Runpod: on-demand Pods Community and Secure rates from runpod.io/pricing, re-read November 5, 2026 ~10:06 PM PT (America/Los_Angeles). The page was marked Updated November 4, 2026. Listed rates matched our prior copy (no change to the overlapping GPUs above).
- Vast.ai: live offers from the public search API (console.vast.ai/api/v0/bundles), read November 5, 2026 ~10:06 PM PT. Filters: rentable, verified, on-demand,
num_gpus=1, standard VRAM band for that card, sorted bydph_total, up to 64 offers. Per-GPU price is the listed total hourly rate (already one GPU in this sample). A100/H100 rows are 80 GB cards. - Machine-readable copy of an earlier same-GPU table: gpu-cost-per-hour.json. Full multi-provider list: gpu-prices.json.
Rent links (referral)
- Rent on Vast.ai Referral link
- Rent on Runpod Referral link
We may earn a commission or account credit if you sign up through those links. Prices are the same either way. Details: disclosure.
Which one should you pick?
- Pick Vast.ai when you will compare hosts and want the lowest on-demand bid (strong tonight on 3090/4090/5090 and A100s).
- Pick Runpod Community when you want one stable published price without shopping hosts (often cheaper than the Vast median).
- Pick Runpod Secure when you want data-center hosting and will pay the Secure premium over Community.
- Size the model first with the VRAM calculator or what can I run so you do not rent more VRAM than you need.