Replit replaces Anthropic with OpenAI as default model provider

Replit’s CEO Amjad Masad announced onstage at TechCrunch Disrupt that the company is no longer working with Anthropic as its default model provider, replacing it with OpenAI. Masad cited performance, cost, and the strength of the partnership as key reasons for the decision. He stated that internal benchmarks and a public benchmark consistently showed OpenAI’s Sol and Astra models performing better for building applications from scratch.

While OpenAI’s ChatGPT is now the default model for Replit’s platform, Anthropic’s Claude models remain available to users. Masad clarified that customers can still access Anthropic models, but the change in default provider is expected to reduce Replit’s spending with Anthropic, as most users tend to stick with the default option.

Prior to this shift, Replit had relied on Anthropic’s Claude models to power its platform, and Masad noted that Replit had been a top-five customer of Anthropic’s until recently. The company describes itself as model-agnostic, allowing users to choose between supported models.

Masad also shared updates on Replit’s revenue growth, noting that enterprise contracts now account for 15% of the company’s revenue this year, up from 2% last year. He expects more than half of Replit’s revenue to come from enterprise and government customers in 2027.

The announcement follows Replit’s August partnership with OpenAI, though no prior indication suggested the end of its Anthropic partnership. Masad’s onstage remarks were disputed by unnamed insiders familiar with the Anthropic-Replit partnership, who claimed the partnership remained active. Public materials, including Replit’s blog and social media posts from late September, still referenced Anthropic’s Claude Fable 5, Claude Fable 5.1, and Opus 5.5 models as part of Replit Agent’s production environment.