Endpoints
OpenAI-compatible endpoints on your cloud or on Inferoptimal's.
Set up and run from Codex
Connect Codex once. Then tell it how many coding agents you want; it picks a measured build, quotes it, launches after you say yes and gives you the one command that adds the endpoint to Codex as its subagents. Your own Codex model stays as it is.
Connecting launches nothing and costs nothing. Codex shows you the price before anything starts.
Put a model on a GPU in the next ten minutes.
Every endpoint has an OpenAI-compatible URL, its own key and live metrics.
Connect a cloud
optional · Inferoptimal cloud needs nonePaste an API key
Get an API key at cloud.lambda.ai. Kept for 30 days; launches start from this console.
Sign in or paste a key
A service-account key with Compute Admin on the project: create one under Service accounts, then Keys, Add key, JSON. Inferoptimal provisions in your project.
Connect a subscription
Find your subscription ID in the Azure portal. You need Owner on the subscription. Inferoptimal holds no key of yours, and you can remove it any time under Service providers in the Azure portal. Azure can take up to 15 minutes to show a new deployment.