The Pirol API
One HTTPS endpoint for language models, embeddings and transcription, run in the EU.
https://endpoint.pirol.ai
It speaks the OpenAI API. If your code already talks to OpenAI, you change the base URL and the key, and nothing else. The quickstart takes you from no account to a first answer.
from openai import OpenAI
client = OpenAI(
base_url="https://endpoint.pirol.ai/v1",
api_key="ep_live_…",
)
What you can do with it
| Page | What it covers |
|---|---|
| Chat completions | Text generation, streaming or not |
| Models | Which models answer, and what they cost |
| Embeddings | Vectors for search and retrieval |
| Audio | Speech to text |
Two ways to pay for the model
The endpoint runs the request either on a model we supply or on your own provider key. Both go through the same URL, and the difference is which key you send — see your own keys.
What is not here
This is the developer documentation for the endpoint. Kyte, the app built on top of it, has its own documentation at kyte.pirol.ai. Routes that only Kyte uses are deliberately absent here.