Replit Shifts Default AI Model From Anthropic to OpenAI, CEO Says
Replit CEO Amjad Masad announces on stage at TechCrunch Disrupt that the AI coding platform is no longer working with Anthropic as its primary provider, citing cost, partnership considerations, and performance. Masad says ChatGPT is now the default, noting that OpenAI's models consistently perform better on internal benchmarks for building vibe-coded applications from scratch. The move follows an OpenAI partnership Replit announced in August.
The shift could significantly reduce Replit's spending with Anthropic, since most customers stick with the default model rather than selecting one themselves. Until recently, Replit had been a top-five customer of Anthropic, relying on Claude models to power its platform, a relationship that helped drive significant revenue growth. Masad clarifies that Replit remains model agnostic and customers can still access Anthropic's Claude models, but OpenAI is now the preferred partner for powering Replit Agent.
However, insiders familiar with the partnership push back on Masad's framing, with one unnamed source saying the Anthropic partnership is still in place. Replit's own website and blog posts promote Anthropic products as recently as late September, including a post citing Claude Fable 5 and Claude Fable 5.1 as models running in Replit Agent's production environment. The full extent of the split between the two companies remains unclear.