Replit makes OpenAI its default for Agent while keeping Claude available
The AMW Read
Replit's preferred-provider switch meaningfully updates its model strategy and highlights coding platforms' influence over supplier demand, although spending and margin effects remain undisclosed.
Named counterparties: Anthropic, OpenAI
Replit makes OpenAI its default for Agent while keeping Claude available
Replit has made OpenAI its preferred partner for powering Replit Agent, CEO Amjad Masad told TechCrunch, citing performance, economics, and the partnership. The clarification narrows his earlier statement that Replit was no longer working with Anthropic: customers can still select Claude and other supported models. Replit announced its OpenAI partnership in August. Masad said internal and public benchmarks informed the decision, but the report provides no benchmark results or pricing terms to independently assess the comparison.
The shift highlights how AI coding platforms can influence model-provider demand through default settings while retaining multiple suppliers. Replit had been a top-five Anthropic customer until recently, according to Masad, making the preferred-provider change commercially relevant even without a complete separation. This also complicates AMW's October 17 coverage of the disputed Anthropic relationship: the clearer development is a change in default and preferred partner. For coding platforms, application-building workflows, deployment, and orchestration may offer differentiation even as underlying model suppliers change; this announcement does not establish a general performance lead for OpenAI.
Builders evaluating Replit should distinguish model availability from the model selected by default and test their own application-building tasks against cost and reliability requirements. For investors, the concrete question is whether the supplier change improves delivery economics as Replit's customer mix shifts. Masad said enterprise contracts now account for roughly 15% of revenue, up from about 2% last year, while B2B exceeds half. Those figures give the model-selection decision an enterprise context, but the report discloses no resulting margin improvement.
