Replit is no longer working with Anthropic, CEO says
TechCrunch Kirsten Korosec, Lucas Ropek, Rebecca Bellan ● Covered by 3 sources
Replit says it’s stopped working with Anthropic and now defaults to OpenAI. The odd part: Anthropic still seems to be available, so this may be a shift in billing more than a clean break.
Based on reporting by TechCrunch, Kirsten Korosec, Lucas Ropek, Rebecca Bellan — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Replit’s CEO says the company has moved away from Anthropic and made OpenAI its preferred partner for powering Replit Agent. Amjad Masad said the change came down to cost, performance, and the partnership itself, and he was blunt about one thing onstage at TechCrunch Disrupt: “I think ChatGPT is better now.”
That sounds like a clean break, but the picture is messier. After the event, Masad clarified that customers can still use Anthropic models inside Replit. OpenAI’s ChatGPT is simply the default now, which matters because most users stick with defaults instead of digging around to change them. In practice, that can cut Replit’s spending on Anthropic without removing it entirely from the product.
Before the OpenAI deal in August, Replit had relied on Anthropic’s Claude models to run the platform. Masad said Replit had recently been one of Anthropic’s top-five customers, and he said the company uses its own internal benchmark, plus a public one, to compare models on performance. He added that Sol and Astra do better on building vibe-coded apps from scratch.
And yet the partnership story isn’t settled. People familiar with the Anthropic-Replit relationship pushed back on Masad’s onstage comments, and Replit’s own website, blog posts, and social media still showed Anthropic models in use as recently as a few weeks ago. A late-September blog post cited Claude Fable 5 and Claude Fable 5.1 in Replit Agent’s production environment, while a Replit product employee also posted about Anthropic’s Opus 5.5 around the same time.
Masad also used the stage time to sketch out how Replit’s business is changing. He said enterprise was about 2% of revenue last year, about 15% this year, and that more than half of revenue is now B2B. Backstage, he said he expects more than half of revenue to come from enterprise and government customers in 2027. He also argued that AI agents need a proper “habitat” with orchestration, safety, sandboxing, networking, identity, and authorization before they can be deployed safely.
My take — AI-written commentary, not fact-checked reporting
This is the usual AI-platform move: keep every model on the shelf, then quietly make one the default and let gravity do the rest. OpenAI gets the front door, Anthropic gets the option menu, and everyone pretends that’s model neutrality. In AI, “vendor agnostic” often just means “we’ll pick first, you can complain later.”
Read more about this at: TechCrunch