How it works
No new syntax to memorize. It's a normal chat until you decide it isn't.
Just start typing
Open Pelladio and type like you would in any chat. No setup, no API key required. Your message goes to a default model and you get an answer.
Pull in another model with @
Type @ and a model's name anywhere in your message (@gemini, @gpt, @grok, @deepseek) and that message routes to them instead of your default. Everyone still reads the same thread, so you never re-explain the problem.
@gemini can you double-check this component for accessibility issues?
Chain several together
Name more than one model in a single message and Pelladio runs them as a pipeline: each one sees what the last one said and builds on it. This is a Pro feature (or free with your own OpenRouter key).
@sonnet write the migration script and @gpt review it for edge cases
Work with real files
Type # and start typing a filename to pull it into the conversation. A real terminal sits in the same window, and edits or commands wait for your okay before anything actually changes.
See it as you build
Working on a web project? A live preview sits right beside the chat, so you watch changes land without switching windows or refreshing a browser tab yourself.
Managed access or your own key, you pick
By default, Pelladio routes your messages through its own connection to every model, so there's nothing to configure. If you'd rather use your own OpenRouter API key, add it in Settings and Pelladio uses that instead: your prompts then go straight from your machine to the model provider, with no ceiling on usage and no cost to Pelladio either way. You can switch between the two whenever you want.
Have a specific question about a feature?
Read the FAQ