Replit says it no longer works with Anthropic, with OpenAI now the default

Masad made the remarks onstage at TechCrunch Disrupt on Thursday, saying the decision rested on several reasons, including cost. He said Replit's internal and public benchmarks show OpenAI's models Sol and Astra performing better at building vibe-coded applications from scratch, and he called OpenAI the preferred partner for powering Replit Agent.

After the event, Masad clarified that users can still choose Anthropic's Claude or other supported models, but ChatGPT is now the default. Because most customers keep the default, the change could significantly reduce Replit's spending with Anthropic, which had until recently been one of its top five customers.

A person familiar with the partnership, who asked not to be named, told TechCrunch that it was still in place. Replit blog posts from late September cited Claude Fable 5 and Claude Fable 5.1 running in Replit Agent's production environment, and a product employee had recently promoted Opus 5.5 on X.

Developers building agents on Replit may find that the default model now routes to OpenAI, which can change output behavior and cost unless they explicitly select a model. Anyone who relies on Claude there should check which model their agents actually call. Model choices made at the platform level can shift even when the underlying tools stay the same.