Replit's CEO says OpenAI has replaced Anthropic as the default model provider for the platform's AI agent. The claim comes from TechCrunch's report on remarks Amjad Masad made onstage at TechCrunch Disrupt.
What Replit says changed
Masad said Replit is no longer working with Anthropic and gave several reasons: cost, partnership and performance. He said an internal benchmark and a public one both show models he called Sol and Astra doing better at building applications from scratch. Replit announced a partnership with OpenAI in August.
After the event, Masad clarified his position. Customers can still pick Anthropic's Claude or other supported models on the platform, but OpenAI is now the default. In his words, Replit "remains model agnostic" and OpenAI is its preferred partner for powering Replit Agent.
The default matters because, as TechCrunch notes, most customers stick with it rather than choosing a model. Masad said Replit was until recently a top-five Anthropic customer, so the move could cut Replit's spend with Anthropic considerably.
What is disputed
The account is not fully settled. A person familiar with the Anthropic and Replit relationship, who was not named, said the partnership is still in place. Replit also promoted Anthropic products recently: a late-September blog post named two Claude models as running in Replit Agent's production environment, and a Replit product employee posted about using Anthropic's flagship model around the same time. Treat the benchmark claims as Replit's own, since the source gives no independent results.
Other figures from the talk
Masad said enterprise contracts were about 2% of Replit's revenue last year and about 15% this year, with more than 50% of revenue coming from B2B. He told TechCrunch he expects more than half of 2027 revenue to come from enterprise and government customers. Replit has not shared ARR figures.
What to do with it
Nothing here requires an urgent migration. The practical step is to confirm which model your Replit projects run on and to choose one on purpose where the output matters.
Top comments (0)