Replit makes OpenAI its default over Anthropic, CEO says
Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.
OpenAI's models now outperform Anthropic's on Replit's benchmarks for code generation, leading to a default switch that will shift most user traffic—previously dominated by Anthropic as their top-5 customer. Engineers relying on Replit’s AI coding tools must expect optimizations and features to increasingly favor OpenAI’s stack, with potential deprecation of Anthropic integrations over time unless explicitly opting out.
Replit has switched its default LLM powering Replit Agent from Anthropic to OpenAI, citing lower costs and superior benchmark performance on generating applications from scratch. Because the vast majority of production agent execution stays on platform defaults, this shifts a top-five Anthropic customer's massive workload directly to OpenAI. For engineers running autonomous coding agents in production, this signals that OpenAI is aggressively closing the performance gap while undercutting on unit economics, making a re-evaluation of model defaults and routing strategies essential for controlling agentic inference spend.