Replit drops Anthropic, switches to OpenAI
Original: Replit is no longer working with Anthropic, CEO says
Why This Matters
Replit's model allegiance signals shifting leverage between OpenAI and Anthropic for dev-tool deals.
Replit CEO Amjad Masad said at TechCrunch Disrupt on Oct. 16 that the AI coding platform ended its Anthropic partnership, citing cost, performance, and partnership fit, having announced an OpenAI deal in August.
Replit CEO Amjad Masad disclosed onstage at TechCrunch Disrupt that the company is no longer working with Anthropic, citing cost, performance, and partnership alignment as the reasons. Replit had announced a deal with OpenAI in August, but gave no public signal at the time that its Anthropic relationship was ending. Masad credited OpenAI's models — specifically referencing Sol and Astra — as outperforming rivals on Replit's internal benchmark for vibe-coded apps built from scratch. "I think ChatGPT is better now," he said flatly.
The statement contradicts sources familiar with the partnership, however. At least one insider, speaking anonymously, told TechCrunch the Anthropic arrangement remains in place. Public evidence backs that up: a late-September blog post names Claude Fable 5 and Claude Fable 5.1 as active models in Replit Agent's production environment, and a Replit product employee promoted Opus 5.5 on X around the same period. TechCrunch has reached out to Replit for clarification.
On business metrics, Masad said enterprise has grown from roughly 2% of revenue last year to about 15% this year, with B2B now accounting for more than 50% of total revenue. He declined to share ARR figures.