OPTES Cloud · Serverless Launch offer
Serverless
Deploy an inference endpoint on a GPU that wakes on the first request and goes back to zero. You pay for request seconds only.
Needs an OPTES account: the same one you use in the app.
How it works
- Sleeps at zeroNo requests, no GPU cost.
- Wakes per requestA worker starts on demand; fast boot keeps the wait short.
- GPU prioritiesOrder the GPUs; we try the first, then the next if it is not available.
- OPTES keyCall the endpoint with your account key; every request is logged.
Prices
Per running worker, per hour and per second.
| GPU | Memory | Per hour | Per second |
|---|---|---|---|
| RTX 2000 Ada | 16 GB | $0.75 | $0.000208 |
| RTX A4000 | 16 GB | $0.75 | $0.000208 |
| RTX A4500 | 20 GB | $0.75 | $0.000208 |
| RTX 4000 Ada | 20 GB | $0.75 | $0.000208 |
| RTX A5000 | 24 GB | $0.97 | $0.000269 |
| RTX 3090 | 24 GB | $0.97 | $0.000269 |
| L4 | 24 GB | $0.97 | $0.000269 |
| PRO 6000 MIG 24GB | 24 GB | $0.97 | $0.000269 |
| RTX 4090 | 24 GB | $1.74 | $0.000483 |
| RTX PRO 4500 | 32 GB | $1.49 | $0.000414 |
| RTX PRO 4500 SE | 32 GB | $1.49 | $0.000414 |
| RTX 5090 | 32 GB | $2.45 | $0.000681 |
| A40 | 48 GB | $1.58 | $0.000439 |
| RTX A6000 | 48 GB | $1.58 | $0.000439 |
| L40 | 48 GB | $2.45 | $0.000681 |
| L40S | 48 GB | $2.45 | $0.000681 |
| RTX 6000 Ada | 48 GB | $2.45 | $0.000681 |
| PRO 6000 MIG 48GB | 48 GB | $2.45 | $0.000681 |
| A100 PCIe | 80 GB | $3.88 | $0.001078 |
| A100 SXM | 80 GB | $3.88 | $0.001078 |
| H100 PCIe | 80 GB | $7.64 | $0.002122 |
| H100 SXM | 80 GB | $7.64 | $0.002122 |
| H100 NVL | 94 GB | $7.64 | $0.002122 |
| RTX PRO 6000 | 96 GB | $4.79 | $0.001331 |
| RTX PRO 6000 WK | 96 GB | $4.79 | $0.001331 |
| RTX PRO 6000 MaxQ | 96 GB | $4.79 | $0.001331 |
| H200 SXM | 141 GB | $9.07 | $0.002519 |
| B200 | 180 GB | $14.01 | $0.003892 |
| B300 | 288 GB | $15.57 | $0.004325 |
Prices read on 2026-11-14, computed from the live catalogue; they can move with the market. The price the console shows before you start is the one billed.
What you can run
- A language-model server with a chat-completions compatible API (vLLM).
- Image generation through ComfyUI.
- Speech to text.
- Any Python worker of your own.
GPU prices are in US dollars, deducted per second from your OPTES Cloud wallet (separate from your chat-plan balance). Top it up in US dollars with cryptocurrency.
Needs an OPTES account: the same one you use in the app.