Replit CEO and co-founder Amjad Masad said onstage at TechCrunch Disrupt on Thursday that the company is no longer working with Anthropic. He cited cost, partnership and performance, and said plainly: "I think ChatGPT is better now."
Masad pointed to Replit's internal benchmark as well as a public one, saying the models it now uses "consistently perform better on building vibe-coded applications from scratch."
The claim is contested. TechCrunch reports that people familiar with the arrangement said the partnership was still in place, and that Replit's own blog posts and social media from late September referenced Anthropic models running in Replit Agent's production environment.
Masad also gave unusually concrete growth numbers: enterprise was about 2% of revenue last year and around 15% this year, with more than half of revenue now B2B. He did not share ARR.
Why it matters
- Model choice is increasingly a cost and benchmark decision, not a loyalty one. If Replit can swap providers this visibly, so can most products built on top of a single lab.
- The dispute itself is instructive: a public statement from a CEO and pushback from people close to the deal, with no clarification yet from the company.
- For anyone building on an API, the practical lesson is that the abstraction layer matters more than the model vendor.