Serverless GPUs for Python
GPUs that wake when
your players do
Put a model behind an API or send a function to a GPU with one decorator. It scales to zero between sessions, and you pay for the seconds it runs.
pip install klio-computeMoving existing GPU code? Change one import
Built for studios shipping AI into games
- 01 / models
One command to a live model
klio model deploy qwen3-14bgives you an OpenAI-compatible endpoint on your own address, protected by your key. - 02 / scale
Zero when nobody's playing
Endpoints sleep when traffic stops and wake on the next request. Keep-warm holds them ready for launch day.
- 03 / migrate
Your existing code, unchanged
klio migrate --checklists what works before you change anything. Most apps need only the import line. - 04 / batch
Fan out in one line
.map()over sprites, transcripts or embeddings in parallel, with results back in input order. - 05 / isolate
Your own kernel
Every customer's GPU work runs sandboxed, apart from other customers and from the machine.
- 06 / bill
Prepaid, to the second
Top up, set a monthly cap, and work stops at zero instead of turning into a surprise invoice.
23 GPU types.
Billed per second.
From a 16 GB A16 to a 288 GB B300. CPU and memory are billed on top, and nothing is billed while an app is scaled to zero.
GPU prices follow the market and are checked every few minutes. A running container keeps the price it started at.
| GPU | Memory | Price per hour |
|---|---|---|
| L4 | 24 GB | $1.45/h |
| L40S | 48 GB | $1.90/h |
| A100-80 | 80 GB | $2.50/h |
| H100 | 80 GB | $3.90/h |
| B200 | 180 GB | Not available right now |
Prices updated
See all 23
| GPU | Memory | Price per hour |
|---|---|---|
| A6000 | 48 GB | $0.75/h |
| RTX4090 | 24 GB | $0.90/h |
| A4000 | 16 GB | $1.20/h |
| L40 | 48 GB | $1.35/h |
| L4 | 24 GB | $1.45/h |
| RTX6000Ada | 48 GB | $1.50/h |
| L40S | 48 GB | $1.90/h |
| A10 | 24 GB | $1.95/h |
| A100 | 40 GB | $2.10/h |
| A5000 | 24 GB | $2.15/h |
| A100-80 | 80 GB | $2.50/h |
| A40 | 48 GB | $2.79/h |
| RTXPro6000 | 96 GB | $3.30/h |
| V100 | 16 GB | $3.55/h |
| V100-32 | 32 GB | $3.55/h |
| H100 | 80 GB | $3.90/h |
| H200 | 141 GB | $6.95/h |
| A16 | 16 GB | Not available right now |
| A30 | 24 GB | Not available right now |
| B200 | 180 GB | Not available right now |
| B300 | 288 GB | Not available right now |
| RTX4000Ada | 20 GB | Not available right now |
| RTX6000 | 24 GB | Not available right now |
# npc_dialogue.py
import klio_compute.compat as klio
app = klio.App("npc-dialogue")
@app.function(gpu="L40S", min_containers=1)
def reply(line: str) -> str:
return model.generate(line)
$ klio deploy npc_dialogue.py
Deployed npc-dialogue: 1 deployment
npc-dialogue-reply