Replit chooses OpenAI as its preferred option over Anthropic, says the CEO.

Replit chooses OpenAI as its preferred option over Anthropic, says the CEO.

Replit, the AI coding platform favored by startup founders and developers alike, has ceased its collaboration with Anthropic, revealed the company’s CEO and co-founder Amjad Masad on Thursday during his appearance at TechCrunch Disrupt.

Masad indicated that the choice stemmed from various factors, cost being one of them. While Replit announced a collaboration with OpenAI in August, there had been no signal indicating a termination of the partnership with Anthropic. 

“I believe ChatGPT is superior now,” Masad remarked, noting that the company employs an internal benchmark to evaluate metrics like performance. “We also release a public benchmark, and consistently, Sol and Astra achieve better results in constructing vibe-coded applications from the ground up. This was a decision influenced by cost, partnership, and performance. For all these reasons, we felt they were the right partner at this time.”

Nonetheless, it appears that Anthropic has not been entirely removed from Replit, which offers a cloud-based software development environment where users can create and launch apps directly in their browsers. 

Following Disrupt, Masad clarified that users still have access to Anthropic models on the platform, but OpenAI’s ChatGPT is now the primary option. This change might lead to a notable decrease in Replit’s expenses with Anthropic, as most users tend to stick with the default setting instead of opting for a different model.

Before establishing a partnership with OpenAI, Replit depended on Anthropic’s Claude models for its platform operations. This collaboration was believed to have significantly contributed to the company’s revenue growth, and Masad remarked during the event that Replit had, until recently, been one of Anthropic’s top five clients.

“Replit continues to be model agnostic,” Masad communicated to TechCrunch over the weekend. “Our users have the option to select Anthropic’s Claude or other supported models…OpenAI is now our preferred partner for powering Replit Agent, based on performance, economics, and the robustness of our partnership.”

However, insiders acquainted with the relationship between Anthropic and Replit disputed Masad’s statements made on stage. One individual familiar with the situation, who opted to remain anonymous, asserted that the partnership remains active.

Evidence from the company’s website, blog entries, and social media updates shows that Replit was still advocating for Anthropic products just weeks ago. For example, a blog post from late September references Anthropic’s Claude Fable 5 and Claude Fable 5.1 as models operational within Replit Agent’s production framework. Simultaneously, a Replit product employee shared on X about utilizing Anthropic’s leading model Opus 5.5 and highlighting its features.

Masad also offered perspectives on the company’s growth while on stage at Disrupt, indicating that enterprise contributions are increasingly becoming a larger segment of revenue.

“We’re not disclosing ARR numbers just yet, but in terms of growth, enterprise was about 2% of revenue last year,” Masad stated. “This year, I would estimate that approximately 15% of our revenue comes from enterprise contracts, with more than 50% of our revenue attributed to B2B.”

Backstage, Masad conveyed to TechCrunch that he anticipates more than half of Replit’s revenue will originate from enterprise and government clients by 2027.

Masad also emphasized the importance of establishing robust AI safety protocols.

“Just as humans require a conducive environment for work—such as a comfortable office, good food, air conditioning, and computers—agents need a suitable habitat to thrive and perform effectively,” he noted. “This habitat must include proper orchestration, effective error recovery, strong safety and security measures, sandboxing, solid networking, and reliable identity and authorization. Once you navigate all of that, you can implement AI safely.”

This article was initially published on October 16 at 2:22 pm PT. It has been revised to provide additional context regarding Replit’s associations with Anthropic and OpenAI. The title has also been updated.

When you make a purchase through links in our articles, we may earn a small commission. This does not impact our editorial independence.