Replit changes its default, not its menu

Replit, the browser-based coding platform, is no longer working with Anthropic, its chief executive and co-founder Amjad Masad said on stage at TechCrunch Disrupt on Thursday, TechCrunch reported. After the event he narrowed the statement: customers can still use Anthropic's models on the platform, but OpenAI's are now the default. TechCrunch first published the story under the headline that Replit was no longer working with Anthropic and later retitled it to say that Replit had made OpenAI its default.

Masad gave three reasons on stage. "I think ChatGPT is better now," he said, adding that Replit uses an internal benchmark and publishes a public one. "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." The benchmark results are Replit's own and were not independently verified.

His clarification, given to TechCrunch over the weekend, was more careful. "Replit remains model agnostic," Masad said. "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."

Replit's own documentation supports the narrower version. Its model selector page lists, for the Power mode, OpenAI's GPT-6 Sol and GPT-6 Luna Fast alongside Anthropic's Claude Sonnet 5 and Claude Sonnet 4.6, and for the Max mode, OpenAI's GPT-6 Astra alongside five Claude models, including Claude Opus 5.5 and Claude Fable 5.1, and Kimi K3. The page does not say which model is the default in either mode. A free tier "always uses intelligent model routing", it says.

The shift has been visible for two months. Replit introduced its free tier on 18 August in a blog post by Masad and Michele Catasta subtitled "With OpenAI, Replit now helps you accomplish even more, every day", running on OpenAI's GPT-5.6 Luna. But a Replit engineering post on 29 September still cited Anthropic's Claude Fable 5 and Fable 5.1 as models running in Replit Agent's production environment, and TechCrunch said a Replit product employee had promoted Anthropic's Opus 5.5 on X around the same time.

Anthropic's side of the account is thin. TechCrunch's report carries no on-record comment from the company. It quotes one person familiar with the partnership, who asked not to be named, as saying the partnership was still in place, and says Masad's on-stage comments "were rejected by insiders". Masad said on stage that Replit had been, until recently, one of Anthropic's top five customers; that ranking is his and has not been confirmed by Anthropic.

What changes in practice is where the default traffic goes. TechCrunch noted that most customers stick with the default rather than choosing a model, so the switch could significantly reduce Replit's spending with Anthropic. Replit has not published usage by model, and neither company has given a figure for what the shift is worth.

Masad also described Replit's revenue mix, though not its size. "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." He told TechCrunch he expects more than half of revenue to come from enterprise and government customers in 2027. For a platform that sells access to other companies' models, the default is a pricing decision as much as a technical one, and Replit has kept the option to reverse it.

Komposite has tracked the price war among model providers, including Google's half-price launch of Gemini 4 Argon for developers and Meta's pricing of its first frontier model. We also reported that OpenAI told investors its run rate was near $50bn.