Replit Shifts Default AI Model to OpenAI Amid Cost Concerns

Replit has transitioned its AI coding platform to feature OpenAI as the default model provider, according to Amjad Masad, the chief executive and co-founder of the company. Speaking at TechCrunch Disrupt, Masad cited better performance metrics and more favorable economics as primary drivers for the switch. The decision marks a significant change for the development environment, which previously relied heavily on Anthropic’s Claude models.

Performance and Economic Benchmarking

Masad noted that internal company benchmarks demonstrated that OpenAI’s ChatGPT offered superior performance for constructing applications from scratch. The Scale Report understands this move serves as a strategic realignment of the platform’s core infrastructure. While Replit remains technically model-agnostic, the shift to OpenAI as the default is expected to substantially reduce the company's previous reliance on its former high-volume contract with Anthropic.

The Future of Enterprise Growth

Despite the pivot, Masad clarified that users retain the ability to select alternative models, including those from Anthropic. The decision comes as Replit experiences a shift in its customer base, with enterprise contracts growing from a marginal 2% of revenue last year to 15% today. Looking ahead, the company expects B2B and government-sector income to represent more than half of its total revenue by 2027.

Maintaining AI Ecosystems

Beyond model selection, Masad emphasized the importance of the technical environment where these agents operate. He described a necessary habitat for AI, focusing on orchestration, sandboxing, and security measures to ensure that autonomous agents can function reliably. This focus on infrastructure is central to the company’s ongoing strategy to provide a secure environment for enterprise software deployment.