The numbers come first. On July 30, OpenAI cut the price of its GPT-5.6 Luna model for API developers by 80%. Less than three weeks later, Replit turned that cost reduction into a product feature — Free Mode — now the default experience for subscribers on its Core ($20/month) and Pro ($100/month) tiers.
Replit's president and head of AI, Michele Catasta, called the approach 'radical.' The logic is straightforward: remove the anxiety of burning through a limited token budget on routine tasks, and reserve the heavier, more expensive models for work that actually demands frontier-level intelligence. When Replit's agent detects a task requires more sophistication, it automatically re-routes the request, then falls back to Free Mode as soon as the task is complete.
'A lot of tasks don't require the frontier-level intelligence,' Catasta said. 'It just requires smaller models that are faster and more affordable.' He described GPT-5.6 Luna as 'very powerful and reliable.'
The efficiency gains behind the price cut came from how the model is run, not from an influx of new compute supply, according to Thibault Sottiaux, Head of Core Products and Platform at OpenAI. 'Month after month we push on the amount of utility you can get for the same dollar amount,' Sottiaux said. 'It's all about using the right tool and the right model for the right outcome.'
The two companies have a long history. OpenAI noted that Replit was an 'early user of GPT-3,' which launched in 2020. Replit itself was founded in 2016 by CEO Amjad Masad. Both Catasta and Sottiaux described a shared vision: making powerful, affordable AI accessible to people without a technical background — the so-called vibe coding demographic.
OpenAI CEO Sam Altman framed the ambition in expansive terms. 'If we can get to a world where anyone with access to the internet can build a product, build a startup, and just kind of get started and get going, I think we're going to see another renaissance-level entrepreneurial boom like we've never seen before,' Altman said in a statement tied to the Free Mode announcement.
Catasta signaled the partnership is only beginning. 'All I'm going to say is that we're working more and more closely with OpenAI, and we're really excited about it. This is the first of many launches we're going to be doing together.'
The market signal here is hard to miss. American AI model builders are competing aggressively on price — partly because enterprise customers are demanding clearer return on investment, and partly because cheap Chinese AI models are forcing the issue. An 80% cost reduction passed directly to end users within weeks is exactly how free enterprise is supposed to work: competition compresses margins, efficiency replaces subsidy, and the product gets better for the person paying the bill. No regulatory mandate required.
