Compare GPU and model prices.
Public prices and marketplace offer data. Not quotes or completed transactions.
Compare a GPU.
Loading comparison…
Average of provider minimums
Waiting for price data.
How this average works
One lowest listed price per provider, with equal weight. Each GPU variant has its own average. Stale sources, partial counts, and occupied prices are excluded.
This describes price levels. It does not match regions, node sizes, or service terms. It is not a quote or a routing recommendation.
| GPU / source | Listed | Occupied | Source details |
|---|---|---|---|
| Loading… | |||
USD per GPU-hour. Marketplace prices are medians. Occupied offers are not proof of completed rentals. Filters apply to this comparison.
Region, node size, fees, and terms can differ. Confirm availability with the provider.
| GPU | Memory | Community | Secure | Stock signal |
|---|---|---|---|---|
| Loading… | ||||
Community: third-party hosts. Secure: vetted data centers. The stock signal covers the lowest-price offer, not each tier.
| GPU | Taken | Taken median | 7-day trend | Open | Open median |
|---|---|---|---|---|---|
| Loading… | |||||
Taken: offers not open to rent. This can include offline machines. These are offer prices, not transaction records.
| GPU | Occupied offers | Occupied median | 7-day trend | Listed | Listed median | RunPod community |
|---|---|---|---|---|---|---|
| Loading… | ||||||
QuickPod marks these offers as not rentable. Occupancy is inferred. Prices come from offers, not invoices.
| Provider | Input | Output | Precision | Context | Uptime |
|---|---|---|---|---|---|
| Loading… | |||||
| Model | Input | Output | Context |
|---|---|---|---|
| Loading… | |||
Sources and freshness.
This page checks for updates every 10 minutes while visible. RunPod, QuickPod, and OpenRouter use 10-minute server caches. Vast.ai uses a 30-minute cache. A refresh can return the previous value while new data loads.
Times show successful server fetches. A stale label marks a failed refresh. Missing sources remain unavailable.
- RunPod pricing: public API list rates. Tier availability is not confirmed separately.
- Vast.ai marketplace: public offer search. Open and taken medians use offer prices divided by GPU count. Partial counts are marked.
- QuickPod open offers and not-rentable offers: medians use offer prices divided by GPU count.
- OpenRouter models: API list rates per million tokens. Provider uptime and precision are reported by OpenRouter.
Cut your GPU bill, or earn from idle GPUs.
Six taps to a plan.