Replit makes OpenAI the default for Replit Agent

The model picker is a footnote. The default is the product.

Replit CEO Amjad Masad said onstage at TechCrunch Disrupt on Thursday that Replit is no longer working with Anthropic, TechCrunch reported. He later clarified to TechCrunch that "Replit remains model agnostic" and that "OpenAI is now our preferred partner for powering Replit Agent." Claude stays available to Replit customers.

Cost, partnership and a benchmark

Masad gave three reasons. "It was a decision based on both cost, partnership, and performance," he said, according to TechCrunch.

On performance, he was blunt: "I think ChatGPT is better now." He said OpenAI's models did better at building vibe-coded apps from scratch, on Replit's internal and public benchmarks.

Masad said Replit was until recently a top-five Anthropic customer. TechCrunch says insiders familiar with the partnership, whose names it withheld, dispute his account, and one says the partnership remains in place. The article carries no statement from Anthropic.

Replit's own record shows how recent the change is. A Sept. 29 Replit post says Replit Agent ran Anthropic's Fable 5 and Fable 5.1 in production, while OpenAI's GPT-6 Astra drove the agent's highest-quality Max mode. In that post, Replit reported Astra at 74% on DeepSWE at $4.43 a task on its top effort setting, and its own agent at 72% for $2.11.

Why the default decides it

Here's the thing. TechCrunch notes that most customers keep the default, so Replit's spending with Anthropic could fall sharply even with Claude still on the menu.

One of the best-known vibe coding platforms is moving its everyday traffic from one lab to another, and its CEO names price as one reason.

Masad also said enterprise grew from about 2% of Replit's revenue last year to about 15% this year, and that more than half of revenue is now business to business. Those customers will notice a change of model before they notice a new feature.

What it means if you build on Replit

The model behind your agent may have changed under you. If you never touched the selector, assume it's OpenAI now.

Rerun the prompts and flows you depend on. Check that the apps it builds still pass your own tests. If Claude did better for your work, choose it in the model selector.

If you build a product on someone else's models, take the lesson for yourself. Partners move. Keep your evals, and your ability to switch, in your own hands.

What to watch

Watch for the new default in Replit's docs. Watch for Anthropic's side of the story.

Then check which model wrote your last deploy.