Replit CEO Amjad Masad announced Thursday at TechCrunch Disrupt that the AI coding platform is no longer partnering with Anthropic, citing cost, partnership quality, and performance as reasons for the shift. Masad said Replit now considers OpenAI's models superior, particularly for building applications from scratch, and the company uses internal benchmarks to assess performance metrics.
The partnership with Anthropic previously enabled Replit to integrate Claude models into its cloud-based development platform, where users write and deploy applications directly in a browser. Masad attributed significant revenue growth to that collaboration.
However, sources familiar with the Anthropic-Replit partnership disputed Masad's claim, with one person requesting anonymity stating the partnership remains active. Evidence on Replit's website and social media supports the disputed account. A blog post from late September cited Anthropic's Claude models, specifically Claude Fable 5 and 5.1, as running in Replit Agent's production environment. A Replit product employee posted on X around the same time promoting Anthropic's Opus 5.5 model and its capabilities.
TechCrunch requested clarification from Replit but had not received a response at publication.
Masad also disclosed revenue shifts at Replit during his Disrupt appearance. Enterprise revenue represented 2 percent of total revenue last year but grew to 15 percent this year. Business-to-business revenue now accounts for more than 50 percent of overall revenue, though Masad declined to share annual recurring revenue figures.
The CEO addressed AI safety requirements separately, describing the infrastructure needed for AI agents to operate safely. Masad stated that agents require proper orchestration, error recovery, security, sandboxing, networking, and identity and authorization systems before deployment.
Replit announced its partnership with OpenAI in August 2024, though the company gave no public indication at that time of ending its Anthropic relationship.