Replit Switches Default AI Model Partner to OpenAI Over Anthropic
Replit has made OpenAI its default partner for Replit Agent, shifting away from Anthropic due to performance, economics, and partnership strength.
Replit CEO Amjad Masad announced at the TechCrunch Disrupt event that the AI coding platform has made OpenAI its default partner over Anthropic. The shift makes OpenAI's ChatGPT the primary engine powering the Replit Agent tool, though customers can still choose to use Anthropic's Claude models manually.
Why did Replit switch its preferred AI provider?
Replit chose OpenAI based on a combination of cost, performance, and partnership dynamics. The company uses both public and internal benchmarks to measure how well models build applications from scratch. Masad stated that ChatGPT is currently performing better for building vibe-coded applications on their platform.
While Replit previously relied heavily on Anthropic's Claude models to power its software development environment, switching defaults significantly changes model usage. Since most users stick to the default setting, the decision is expected to reduce Replit's overall spending with Anthropic.
Is Replit still working with Anthropic?
Replit remains a model-agnostic platform. Users retain the ability to select Anthropic models, such as Claude, inside the cloud-based development environment. Masad clarified that while OpenAI is now the preferred partner for Replit Agent, Anthropic's technology is not completely removed from the system.
Sources familiar with the partnership noted that the relationship remains in place, and Replit recently promoted Anthropic models in production environments and social media posts. Until recently, Replit was a top-five customer for Anthropic.
What does this mean for business growth and AI deployment?
Replit is seeing a shift toward business clients, with corporate contracts driving an increasing portion of its revenue. Masad shared that enterprise clients accounted for about 2% of revenue last year but have grown to around 15% this year, with overall B2B revenue making up more than 50% of the business.
To support this corporate expansion, Masad highlighted the need for secure AI environments. Safe deployment of AI agents requires underlying infrastructure that provides strong error recovery, security sandboxing, identity authorization, and proper orchestration.
Comments 0
No comments yet. Start the conversation.