A live exchange for GPU compute. Hundreds of vetted providers stream real-time prices, latency and availability into one open marketplace — rent capacity by the second, or sell your idle GPUs and get paid as the work settles.
A live trading floor for GPU compute — real-time prices, latency, and availability streaming across providers and regions, settled by the second.
| Provider | GPU | Region | 12h | Latency | Price | Status |
|---|---|---|---|---|---|---|
| Provider-01 | RTX 4090 | us-ashburn | 34ms | $0.14/hr | Online | |
| Provider-02 | A100 | us-oregon | 41ms | $0.28/hr | Online | |
| Provider-03 | H100 | eu-frankfurt | 26ms | $0.79/hr | Busy | |
| Provider-04 | MI300X | ap-singapore | 37ms | $0.66/hr | Online | |
| Provider-05 | RTX 5090 | us-dallas | 22ms | $0.31/hr | Online |
- ▸Provider-03·H100·rented 4h·$3.16·eu-frankfurt
- ▸Provider-01·RTX 4090·rented 8h·$1.12·us-ashburn
- ▸Provider-04·MI300X·rented 2h·$1.32·ap-singapore
- ▸Provider-02·A100·rented 6h·$1.68·us-oregon
Requests route to the nearest region — low latency, anywhere on earth. Eight edge locations on five continents keep inference close to your users.
Everything you need to know about renting and selling GPU compute on the exchange — pricing, vetting, payouts and uptime.
Pricing is a live, open market. Providers set their own $/hr rate per GPU and the marketplace surfaces the best available price for each model and region in real time. You pay only for what you use, billed by the second — no minimums, no idle charges. As supply and demand shift, prices update on the order book so you always see the true clearing rate.