AI Coding Platform Dumps Anthropic for OpenAI, Says ChatGPT Models Cheaper and Better

Replit's shift to making OpenAI the default provider for Replit Agent, rather than completely removing Anthropic's models, indicates a re-prioritization of its primary AI partner. The change in default status can significantly redirect usage and spending towards OpenAI, even if Claude models remain available as an option. This adjustment is based on Replit's internal evaluations, which reportedly showed OpenAI's models offering better performance and economics for its specific AI-assisted coding tasks.

The initial, more definitive statement by Replit's CEO about ceasing work with Anthropic was later clarified, emphasizing that Replit remains "model agnostic" and users can still choose other models. This distinction is important: it frames the change as a strategic preference rather than a complete termination of the relationship. Evidence from Replit's engineering updates, describing Claude models running in production recently, further supports that the connection with Anthropic has not been entirely severed.

While Replit reported being one of Anthropic's top five customers, the precise financial impact of this change on both companies remains unquantified. The claimed cost efficiencies and performance improvements of OpenAI's models for Replit's workloads are not supported by publicly disclosed pricing figures or independently reproducible benchmark results, making it difficult for external parties to assess the full extent of the advantage.