Replit, the cloud-based coding platform that runs in the browser, announced on Thursday that it now treats OpenAI as its primary partner. The statement was made by Amjad Masad, Replit’s CEO and co-founder, while speaking at TechCrunch Disrupt.
Masad said the change was driven by several factors, including cost and performance. Replit announced a partnership with OpenAI in August; until Masad’s remarks there had been no public indication that the company was stepping back from Anthropic.
“I think ChatGPT is better now,” Masad said on stage, adding that Replit relies on internal benchmarks to evaluate performance. “We publish a public benchmark as well, and consistently, Sol and Astra perform better on building vibe‑coded applications from scratch. It was a decision based on both cost, partnership, and performance. And for all these reasons, we just felt like they were a better partner for now.”
Anthropic remains accessible but OpenAI is the default
After the Disrupt session, Masad clarified that Anthropic models are still accessible to customers on Replit’s platform, but OpenAI’s ChatGPT is now the default model. Because most customers stick with defaults, this change could substantially reduce Replit’s spending with Anthropic.
Before striking the OpenAI deal, Replit relied on Anthropic’s Claude models to power its platform. That prior partnership was reported to have helped drive Replit’s revenue growth; Masad noted on stage that Replit had been, until recently, among Anthropic’s top‑five customers.
Speaking to TechCrunch over the weekend, Masad reiterated: “Replit remains model agnostic. Our users can choose Anthropic’s Claude or other supported models…OpenAI is now our preferred partner for powering Replit Agent, based on performance, economics, and the strength of our partnership.”
Conflicting accounts and public evidence
Some insiders familiar with the Anthropic–Replit relationship pushed back on Masad’s onstage characterization. One person who asked not to be named said the partnership was still in place.
Public materials, including Replit’s website, blog posts, and social media, show the company was still promoting Anthropic products as recently as a few weeks before Masad’s remarks. A late‑September blog post references Anthropic’s Claude Fable 5 and Claude Fable 5.1 as models running in Replit Agent’s production environment. Around the same time, a Replit product employee posted on X about using Anthropic’s flagship model Opus 5.5 and highlighted its capabilities.
Revenue mix and AI safety
Masad also discussed Replit’s growth and shifting revenue composition. “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,” he said. “This year, I would say like 15% of our revenue is enterprise contracts, and more than 50% of our revenue is B2B.”
Backstage, Masad told TechCrunch he expects more than half of Replit’s revenue to come from enterprise and government customers by 2027.
He stressed the need for robust AI safety and operational controls, saying agents require an appropriate ‘‘habitat’’ — orchestration, error recovery, safety and sandboxing, networking, identity and authorization — to be deployed safely.
This article was originally published on October 16, 2024 at 2:22 pm PT and has been updated to include additional context about Replit’s relationships with Anthropic and OpenAI.
