Replit Parts Ways with Anthropic, Citing Cost and Performance in Shift to OpenAI

Replit, the popular cloud-based AI coding platform that has evolved into a staple for startup founders and software developers alike, is no longer working with AI lab Anthropic. The surprising revelation was made by Amjad Masad, the CEO and co-founder of Replit, while speaking onstage at the TechCrunch Disrupt conference.

Addressing an audience of industry professionals and founders, Masad explained that the decision to sever ties with Anthropic was driven by a combination of factors, chief among them being cost, the nature of the partnership, and shifting performance benchmarks. While Replit officially announced a major partnership with rival AI giant OpenAI back in August, there had been no public indication at the time that the platform was quietly moving away from its foundational relationship with Anthropic.

The historical alliance between Replit and Anthropic had previously been touted as a major growth engine for the coding platform. Under that partnership, Replit leveraged Anthropic’s advanced Claude models to power its browser-based development and deployment environment. Industry reports and public analyses have credited that integration as a critical catalyst that helped Replit scale its revenue tenfold following the initial collaboration. However, as the generative AI landscape matures and competition among foundational model providers intensifies, Replit has apparently decided to pivot its primary allegiance.

"I think ChatGPT is better now," Masad told the TechCrunch Disrupt audience, pointing to the continuous evolution of OpenAI’s technology. He elaborated that Replit relies on a rigorous internal benchmarking system to evaluate performance metrics across various models. According to Masad, the company also publishes public benchmarks that consistently demonstrate how OpenAI-powered tools perform superior tasks, particularly when it comes to building complex, vibe-coded applications entirely from scratch.

"It was a decision based on both cost, partnership, and performance," Masad said, summarizing the multifaceted evaluation that led to the split. "And for all these reasons, we just felt like they were a better partner for now."

Despite Masad’s definitive remarks onstage, the status of the relationship has generated conflicting accounts behind the scenes. Following the announcement, insiders familiar with the commercial agreements between Anthropic and Replit pushed back against the narrative of a clean break. One individual with direct knowledge of the situation, who spoke on the condition of anonymity, maintained that the partnership was still active in some capacity.

Furthermore, concrete evidence suggests that Replit was integrating Anthropic products into its ecosystem as recently as September. Documentation from a public Replit blog post published in September explicitly cites Anthropic’s Claude models—specifically Claude Fable 5 and Claude Fable 5.1—as actively running within the production environment of Replit Agent, the platform’s autonomous coding assistant. In response to these conflicting signals, TechCrunch reached out to Replit for formal clarification regarding the timeline and current status of the technical integration.

Beyond the shifting dynamics of its AI model providers, Masad used his appearance at TechCrunch Disrupt to provide a broader window into Replit’s rapid financial and commercial expansion. While the company has opted to keep its exact Annual Recurring Revenue figures under wraps for the time being, the CEO highlighted a dramatic structural shift in where that revenue is coming from, noting that enterprise customers are beginning to command a significantly larger share of the business.

"We’re not sharing ARR numbers just yet, but in terms of growth, like, I would say, enterprise was like 2% of revenue last year," Masad said, illustrating the magnitude of the change. "This year, I would say like 15% of our revenue is enterprise contracts, and more than 50% of our revenue is B2B."

This enterprise growth comes at a time when the broader software development sector is grappling with how to safely and effectively deploy autonomous artificial intelligence into professional workflows. During his discussion, Masad emphasized that the successful integration of AI requires far more than just raw model intelligence; it demands a robust infrastructure and carefully controlled operational environments.

Drawing a parallel to human work environments, the Replit co-founder argued that autonomous software agents require a specialized digital ecosystem to function properly and securely.

"In the same way that us humans need a certain environment to work — you need a nice office, maybe nice good food in the office, air-conditioning, computer — agents need a habitat for them to exist and do good work," Masad said. He noted that building such a habitat involves implementing strong orchestration frameworks, effective error recovery protocols, comprehensive safety, security, sandboxing, advanced networking, and reliable identity and authorization management. According to Masad, once developers and enterprises successfully figure out and establish all of those foundational elements, they can finally deploy AI safely and at scale.