Anthropic pitches AI safety bill setting thresholds for model suspensions
The proposal would establish independent evaluations and a court process for halting frontier AI models, according to a media report.
Replit CEO says Anthropic partnership has ended, but insider disputes claim
Amjad Masad cited cost and performance in favoring OpenAI, while a person familiar with the Anthropic relationship said it remained in place.
Replit has stopped working with Anthropic, CEO and co-founder Amjad Masad said at a technology conference, citing cost, performance and partnership considerations. A person familiar with the relationship disputed his account, saying the partnership remained in place.
The AI coding company announced an OpenAI partnership in August. Masad said Replit considered OpenAI the better partner for now, based partly on assessments of performance. He said the company uses both internal testing and a publicly available benchmark.
Replit provides browser-based tools for developing and deploying applications. Its arrangement with Anthropic involved using the AI developer’s Claude models to power the platform.
A September blog post identified Anthropic’s Claude Fable 5 and Claude Fable 5.1 among the models operating in Replit Agent’s production environment. The person who said the partnership was continuing requested anonymity.
Masad also described a growing contribution from enterprise customers. He estimated that enterprise contracts accounted for about 15% of revenue, compared with about 2% a year earlier, and said business-to-business customers generated more than 50% of revenue. Replit was not yet disclosing annual recurring revenue figures, he said.