Replit has launched FreeMode, a new feature powered by OpenAI’s GPT-5.6 Luna model that is designed to help users get more use out of their AI subscription without burning through tokens on everyday tasks.
The feature, which Replit announced on August 19 as part of its expanding partnership with OpenAI, is now the default for Core and Pro subscribers.
It gives users a way to chat, ideate, and handle simpler tasks with Luna, while more advanced models are reserved for work that requires greater reasoning or coding capabilities.
Louis Eriakha
Is FreeMode actually free?
Despite the name, FreeMode is not a free plan option. Users still have to subscribe to a paid plan on the vibe-coding platform to access FreeMode.
In the startup's announcement, FreeMode is available to users on its Core plan, which costs $20 per month, and $100 per month for the Pro plan. This allows users to "create more for just the cost of their monthly subscription," the company said on its web page.
That means a user can brainstorm an idea or work through everyday tasks without the stress of worrying about burning AI tokens.
How does FreeMode reduce token burn?
The new feature, while built for cost efficiency, uses OpenAI’s GPT-5.6 Luna to handle everyday, lower-intensity tasks instead of relying on more expensive frontier models for every request.
Catasta said the feature became possible after OpenAI cut the cost of running Luna by 80% on July 30.
“A lot of tasks don’t require the frontier-level intelligence,” Catasta told Fortune. “It just requires smaller models that are faster and more affordable.” For that reason, GPT-5.6 Luna is “very powerful and reliable,” he noted.
Subscribe for free to continue reading this article
Subscribe SubscribeAlready have an account? Log in
