Replit, an AI coding platform popular with startup founders and developers, will no longer partner with Anthropic, company CEO and co-founder Amjad Massad said on stage at TechCrunch Disrupt on Thursday.
Massad said the decision was made for a variety of reasons, including cost. Replit announced its partnership with OpenAI in August, but there has been no indication that it will end its partnership with Anthropic.
“I think ChatGPT is better now,” Massad said, adding that the company uses internal benchmarks to evaluate performance and other metrics. “We also publish public benchmarks, and Sol and Astra consistently show superior performance when building vibe-coded applications from scratch. This was a decision based on both cost, partnership, and performance. For all of these reasons, we felt they were the better partner so far.”
However, it appears that Anthropic is not completely separated from Replit, which sells a cloud-based software development platform that allows users to develop and deploy apps directly from within their browser.
Massad revealed that after Disrupt, customers will still have access to Anthropic models on the platform, but OpenAI’s ChatGPT is now the default. This move could significantly reduce Replit’s spending on Anthropic, as most customers will continue to use the defaults rather than choosing a model themselves.
Before signing the deal with OpenAI, Replit relied on Anthropic’s Claude model to run its platform. The partnership is said to have contributed to a significant increase in revenue for the company, with Massad noting on stage that Replit was until recently among Anthropic’s top five customers.
“Replit is still model agnostic,” Massad told TechCrunch over the weekend. “Our users can choose Anthropic’s Claude or any other supported model. Based on performance, economics, and strength of partnership, OpenAI is now our preferred partner to power Replit Agent.”
But Massad’s comments on stage were rejected by insiders familiar with Anthropic’s partnership with Replit. One person familiar with the matter, who asked not to be named, said the partnership is still ongoing.
There is evidence on the company’s website, blog posts, and social media posts that Replit was promoting Anthropic products as recently as several weeks ago. For example, a blog post from late September cited Anthropic’s Claude Fable 5 and Claude Fable 5.1 as models running in Replit Agent’s production environment. Around the same time, one of Replit’s product reps posted on X about using Anthropic’s flagship model Opus 5.5 and promoting its features.
Massad also shared insights about the company’s growth during the disruption stage, noting that Enterprise is starting to deliver greater revenue.
“We haven’t shared the ARR numbers yet, but from a growth perspective, enterprise was around 2% of our revenue last year,” Massad said. “This year, I think 15% of our revenue will be corporate contracts and over 50% of our revenue will be B2B.”
Backstage, Massad told TechCrunch that he expects more than half of Replit’s revenue to come from enterprise and government customers by 2027.
Massad also commented on the need for better AI safety measures.
“Just as we humans need a certain environment to work, we need a nice office, maybe good food in the office, air conditioning, a computer. Agents also need a habitat to exist and do good work,” he said. “So that habitat has the right orchestration, the right error recovery, the right safety and security and sandboxing, the right networking, the right identity and authentication. And once you understand all of that, you can deploy AI securely.”
This article was originally published on October 16th at 2:22pm PT. This has been updated to include more context about Replit’s partnerships with Anthropic and OpenAI.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
