Replit's Anthropic Status Is Disputed, but Its Model Strategy Is Clear

Replit chief executive Amjad Masad said the coding platform was no longer working with Anthropic, citing cost, performance and partnership considerations. The statement would mark a meaningful supplier change for a company whose growth was closely associated with Claude models, but TechCrunch reported conflicting evidence.

A person familiar with the relationship told the publication that the partnership remained in place. Replit's own recent materials had also referenced Anthropic models in production, including a late-September blog post. Those facts do not necessarily contradict Masad if the relationship changed afterward or if older integrations remain available, but Replit had not publicly reconciled them.

Masad said Replit uses internal and public benchmarks and argued that OpenAI models performed better for building applications from scratch. Replit announced an OpenAI partnership in August. Benchmark results can guide model selection, yet they depend on tasks, prompts, cost assumptions, latency and the surrounding product system.

The commercial mechanism is more important than a model leaderboard. AI coding platforms pay inference costs while competing on speed, reliability and the quality of generated applications. Moving volume between suppliers can improve gross margin or product performance, but it can also increase dependence on a new partner.

For Anthropic, the reported change would matter as evidence about customer retention, not as a disclosed financial event. Neither private company has provided contract value, usage volume or revenue exposure, so the financial effect cannot be calculated responsibly.

The immediate conclusion must remain narrow. Replit's CEO described a move away from Anthropic, while a knowledgeable source and recent product evidence disputed a complete break. A platform can also retain technical compatibility after shifting most production traffic, which makes binary partnership labels especially imprecise. Updated model documentation, a clarified company statement or contract disclosure would resolve the question. Until then, the strategic direction toward OpenAI is visible, but the exact Anthropic status is not.