Replit Anthropic Split: CEO Says It's Over, Source Says No

The Replit Anthropic split just became one of the messiest stories in AI coding. Replit chief executive Amjad Masad told a TechCrunch Disrupt crowd this week that his company is no longer working with Anthropic, the maker of Claude, according to TechCrunch. But a person familiar with the relationship says the partnership is still in place.

So which is it? Right now the public record does not settle the question, and that gap is the real story. Here is what is confirmed, what is disputed, and why developers on both sides are paying attention.

What Masad said on stage

Speaking at TechCrunch Disrupt in San Francisco on Thursday, Masad said Replit stopped working with Anthropic. He described the decision as "based on both cost, partnership, and performance," according to the report summarized by ExplainX. He also added a pointed line about quality, saying ChatGPT is better now.

That is a blunt thing for a founder to say about a supplier that sits at the centre of AI coding. Replit sells tools that let people build and ship software with the help of AI agents, so the models underneath matter a lot to its product and to its bills.

Why the story is disputed

TechCrunch reports that an anonymous person familiar with the relationship said the Anthropic partnership remains in place. On top of that, a September Replit engineering post listed Claude models among those running in the production Replit Agent, which suggests Claude was still in use last month.

That September evidence shows Claude was in production then, but it does not prove what the arrangement looked like on October sixteenth. TechCrunch asked Replit to clarify and said it would update its story. As of the latest reports, neither company had issued a statement that resolves the conflict.

Another outlet, Crypto Briefing, framed Masad's remarks more softly. It said Replit has not publicly announced any termination, and that his larger message was about avoiding dependence on any single model provider, because the labs supplying models are increasingly becoming the competition.

The cost and competition angle

Strip away the drama and the business logic is easy to follow. AI coding products burn through huge amounts of model usage, so even small price differences add up quickly. Replit also announced a partnership with OpenAI in August, which means it already had a second major supplier ready to go.

Replit says it wants to assign different models to different jobs, aiming for high quality at the lowest cost rather than using one model for everything. That multi-model approach is becoming the standard playbook for AI app builders who do not want their whole company tied to one lab.

The tension is that model makers now sell their own coding tools. When your supplier is also your competitor, every price increase or product launch can feel like a threat. That is why more startups are building in the ability to swap models quickly.

It is worth remembering how quickly this market moves. Only a year or two ago, most coding assistants leaned on a single model family. Today, new model versions land every few weeks, and benchmark leaders change often. A platform that locks itself into one provider risks falling behind on quality or overpaying for features it does not need, so flexibility has become a selling point in its own right for founders and investors alike.

What it means for people who build with Replit

For everyday users, the practical impact is unclear for now. Whether Claude stays available inside Replit, and how Replit Agent behaves, depends on facts that neither company has laid out publicly. Expect clarification soon, because a headline like this puts pressure on both sides to speak.

There is also a trust question for customers. Companies that build products on top of Replit want to know which models touch their code and data, because each provider has its own terms, pricing and safety rules. A clear public statement from Replit about which providers it uses would help those customers plan, and it would end the guessing game that this week's conflicting reports created.

For the wider industry, the Replit Anthropic split is a reminder that AI partnerships are shifting fast and that public statements can run ahead of the paperwork. Read the claims carefully, watch for official statements, and treat any single account as incomplete until Replit or Anthropic responds. More of our tech coverage will follow as the story develops.