Get started
Introduction
Use Decillion's models, computers and storage from your own code, billed in US dollars from your own balance.
What the API gives you
The Decillion developer API is for people who want our infrastructure without our app: run models of every kind, run commands on clean Linux computers, and keep files, all through a plain HTTPS API.
Service
What it is
Billed by
One catalog of language, image, video, speech, transcription, embedding, decision and rerank models behind one API.
Tokens, images, seconds of video or audio, characters or searches, as each model is priced
Linux machines you create, start, stop and run commands on.
The minute, while running
Files your code uploads, reads and shares with your computers.
To be announced at launch
Base URL
Every request goes to the API host, over HTTPS only:
https://api.decillionai.com/v1
This host serves developer keys only. It is separate from the one the Decillion apps use, so a key can never reach your app account's other features.
How it works
1.
Pick an organization. Your personal account is one; you can make others for a team, each with its own balance, keys and people. See
Organizations.
2.
Create a key in the console. Choose what it may do, and optionally a spend cap, a rate limit, and the models it may use.
3.
Send requests with the key as a bearer token. Every request is attributed to the key that made it.
4.
Watch it in Usage and Logs: spend by model, key and day, and the full request and response of every call.
5.
Pay from the organization's balance. It is prepaid, in US dollars. When it runs out, requests stop; you are never invoiced after the fact.
OpenAI compatibility
The chat endpoint follows the OpenAI Chat Completions format. Any OpenAI SDK works if you change two things: the base URL and the key.
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.DECILLION_API_KEY,
baseURL: "https://api.decillionai.com/v1",
});
const completion = await client.chat.completions.create({
model: "google/gemini-3.8-flash",
messages: [
{
"role": "user",
"content": "Say hello in one sentence."
}
],
});
console.log(completion.choices[0].message.content);