Replit Drops Anthropic, CEO Says OpenAI's Models Now Win

Replit is no longer working with Anthropic, CEO and co-founder Amjad Masad said Thursday onstage at TechCrunch Disrupt. The AI coding platform, which ran on Anthropic's Claude models, made the call for a mix of reasons — cost, partnership terms and performance — even though it never publicly flagged the split after announcing an OpenAI partnership in August.

"I think ChatGPT is better now," Masad said. Replit relies on an internal benchmark to assess performance, plus a public one. "We publish a public benchmark as well, and consistently, Sol and Astra perform better on building vibe-coded applications from scratch," he said. "It was a decision based on both cost, you know, partnership, and performance. And for all these reasons, we just felt like they were a better partner for now."

Is the Anthropic partnership really over?

Not everyone buys it. One person familiar with the Anthropic-Replit relationship, who asked not to remain unnamed, told TechCrunch the partnership was still in place.

There's evidence backing that claim. A Replit blog post from September cites Anthropic's Claude Fable 5 and Claude Fable 5.1 as models running in Replit Agent's production environment. TechCrunch said it reached back out to Replit for clarification and will update its story if it hears back.

What does the model switch mean for revenue?

The Anthropic deal mattered commercially. Replit used Claude models to power its cloud-based platform, where users build and deploy apps directly from a browser. TechCrunch reported that the partnership helped Replit grow its revenue significantly.

Masad wouldn't give an ARR figure. He did break down the mix, and the enterprise shift is the headline. "We're not sharing ARR numbers just yet, but in terms of growth, like, I would say, enterprise was like 2% of revenue last year," Masad said. "This year, I would say like 15% of our revenue is enterprise contracts, and more than 50% of our revenue is B2B."

That's a jump from 2% to roughly 15% of revenue from enterprise contracts in a single year. More than half of revenue now comes from business customers.

How does Replit approach AI safety?

Masad framed agent safety as an infrastructure problem, not a policy debate. Agents, he argued, need the same kind of supportive setup that human workers expect.

"In the same way that us humans need a certain environment to work — you need a nice office, maybe nice good food in the office, air conditioning, computer — agents need a habitat for them to exist and do good work," he said. "So that habitat has good orchestration, good error recovery, good safety and security and sandboxing, good networking, good identity and authorization. And once you figure all of that out, then you can deploy AI safely."

The components he named:

• Orchestration
• Error recovery
• Safety, security and sandboxing
• Networking
• Identity and authorization

The contradiction between Masad's onstage claim and the September production evidence leaves the exact state of the Anthropic relationship unclear. If Replit has fully moved its stack to OpenAI's models, it becomes one of the highest-profile defections from Claude among AI-native developer tools — and a data point in the escalating benchmark-and-price war between frontier labs.