AI tokens.Swisscompute.

One API for Swiss-hosted AI. Let it pick the model for each task, or choose one yourself.

CHF 2.00 of free credit when you confirm your e-mail address. No card, no subscription.

Auto Model Selection

Connect once. With the model auto, the API matches the task to a model, respects your spending limit and keeps ongoing work on a consistent route.

Choose an example

Illustrative decisions at the catalogue's rates. No live inference or prompt analysis.

Example decisionCH / Switzerland

Selected model

Qwen3.6 35B

This example sends structured extraction to the lower-cost model.

Input: 4'000 tokens × CHF 0.10 / M
Output allowance: 1'000 tokens × CHF 0.70 / M
Maximum example charge: CHF 0.00110
Remaining budget: CHF 10.00

Within budget · CHF 0.00110 reserved before generation

Automatic selection is included in the model rates.

See the connection example

Models & pricing

CHF per million tokens.
Input and output billed separately.

ModelInputOutput

Prices exclude VAT. No subscription: you pay what you use, from prepaid credit you top up by card. A request refused before it runs is not charged. With auto, you pay the rates of the model that answers. Every model and its limits

Estimate your cost

Qwen3.8 27B

Type your own numbers: the total changes as you type. Choose the model in the list above.

Requests / monthInput tokens / requestOutput tokens / request

Per month, with the selected modelCHF 2.95

Input CHF 1.20 · Output CHF 1.75

One request: CHF 0.00295

One connection.
Your choice.

Let the API choose the model with auto, or call a model directly. Keep the same API connection.

Set your base URL and API key in your app.
Select auto, or a model from the catalogue.
Set spending limits in your account.

Set a monthly spending limit for each key in your dashboard: it is checked before generation.

Quickstart in the docs

Stream the response

Example only. No request is sent from this page.