Replit picks OpenAI over Anthropic as primary AI provider, says ChatGPT is better now

Popular AI coding platform Replit has made a major change that will impact how developers build software. The company has announced that it is making OpenAI its default artificial intelligence model for its popular coding assistant, known as Replit Agent. This means OpenAI will replace Anthropic's Claude, which was previously the primary engine powering the platform.

The major shift was revealed by Replit CEO Amjad Masad during the recent TechCrunch Disrupt event, where his initial words caused quite a bit of confusion. He first suggested that Replit was completely done working with Anthropic. However, he later clarified that the partnership is still active, even though OpenAI has now taken the main seat.

Why Replit made the switch to OpenAI

Masad explained that the decision to prioritize OpenAI came down to three main factors: better performance, lower costs, and a stronger business partnership. Replit and OpenAI had already signed a close partnership agreement in August 2026, which laid the foundation for this transition.

During his talk, Masad plainly stated that he believes ChatGPT is simply a better tool right now. According to the company's internal testing and public benchmarks, OpenAI's newer models like Sol and Astra do a much better job when it comes to building applications completely from scratch. On top of the performance boost, managing the high costs of running complex AI models played a major role in choosing OpenAI over Anthropic.

Anthropic is not completely out of the picture

Despite Masad's initial comments on stage, insiders and company records confirm that the relationship with Anthropic is not dead. In fact, Replit employees were actively promoting Anthropic products like Claude Fable 5 and its flagship Opus 5.5 model on social media just a few weeks ago. Replit maintains that its platform remains open to different models, and users who prefer Anthropic's Claude can still go into their settings and manually select it for their coding tasks.

However, losing the default spot is a massive setback for Anthropic. In the tech industry, the vast majority of users stick to the default settings and never bother to change them. Because of this, Anthropic is expected to see a huge drop in usage on the platform. This is a significant blow, especially considering Masad noted that Replit was previously one of Anthropic's top five biggest customers.

Massive growth in enterprise business

Aside from the AI model wars, Masad also shared some impressive data about how Replit is growing its business. The company is seeing a massive shift toward larger corporate clients. Last year, enterprise contracts made up a tiny 2% of Replit's total revenue. This year, that number has jumped to 15%.

Overall, more than half of the company's revenue now comes from business-to-business sales rather than individual developers. Looking ahead, Masad expects that by the year 2027, over half of Replit's total earnings will come directly from large enterprise and government customers.

The need for a safe AI environment

Masad also touched upon the importance of AI safety, using a clever comparison to explain how AI needs a good workspace. He noted that just like human workers need a nice office, good food, and air conditioning to do their best work, AI coding agents need a proper "habitat" to function well.

For artificial intelligence, this habitat includes good security, error recovery, and safe networking. He stressed that only when all these safety measures are properly set up can AI be deployed securely for developers to use.