Replit, the coding platform powered by AI, known for its popularity among startup founders and developers, has decided to discontinue its collaboration with Anthropic, as stated by the company’s CEO and co-founder, Amjad Masad, during his appearance at TechCrunch Disrupt.
Masad explained that the decision stemmed from various factors, including cost considerations. Although Replit formed a partnership with OpenAI in August, there had been no prior indication that it would cease its relationship with Anthropic.
“I believe ChatGPT is currently superior,” Masad mentioned, adding that their internal benchmarks assess metrics like performance. “We also publish a public benchmark, and consistently, Sol and Astra excel in creating vibe-coded applications from scratch. This choice was influenced by cost, partnership, and overall performance. For all these reasons, we felt they were a more suitable partner for the time being.”
Replit offers a cloud-based development platform that enables users to build and deploy apps right from their browsers. Through its collaboration with Anthropic, Replit had utilized the AI lab’s Claude models to enhance its services, a partnership that reportedly contributed to a significant increase in the company’s revenue.
However, Masad’s remarks were challenged by sources familiar with the previous partnership between Replit and Anthropic. One individual, who preferred to remain anonymous, claimed that the partnership remains intact.
Evidence on Replit’s website, along with recent blog posts and social media updates, suggests that the company was still utilizing Anthropic products just weeks ago. For instance, a late September blog highlighted the use of Anthropic’s Claude Fable 5 and Claude Fable 5.1 models within Replit Agent’s production environment. Around that same time, a Replit product team member mentioned the capabilities of Anthropic’s flagship model, Opus 5.5, on social media.
TechCrunch followed up with Replit for further clarification and will update the article pending a response.
During his speech at Disrupt, Masad also shared insights into the company’s financial growth, revealing that income from enterprise clients is becoming a more substantial part of their overall revenue.
“While we aren’t disclosing annual recurring revenue figures just yet, I can say that last year, enterprise contributed about 2% to our revenue,” Masad noted. “This year, it’s grown to approximately 15%, while over 50% of our revenue now comes from B2B partnerships.”
Masad emphasized the importance of proper safety measures in AI development.
“Just as humans require a conducive environment to perform effectively — nice offices, quality food, air conditioning, and computers — AI agents also need a suitable habitat to thrive and produce good work,” he explained. “This includes proper orchestration, error recovery, safety and security measures, sandboxing, robust networking, and identity management. Once those elements are in place, AI can be deployed safely.”
This article has been revised to provide additional context regarding Replit’s use of Anthropic’s models.
