Hype check

"Replit Dropped Anthropic": One Sentence From the Stage Against Fresh Evidence

Replit's CEO said on stage the company is done with Anthropic, but a source, the company's own recent posts and the lack of any numbers say the story is less clear.

What was claimed

At TechCrunch Disrupt, Replit CEO Amjad Masad said his company no longer works with Anthropic. His reasons: cost, partnership and performance. He added that "ChatGPT is better now" and that OpenAI is "a better partner for now." Headlines read it as a clean break.

What is actually there

The claim is a spoken statement with no document behind it. Replit gave no figures: no cost comparison, no benchmark scores, no ARR. Masad only said that an internal and a public benchmark favor OpenAI models on building apps from scratch, without naming the tests or results.

Meanwhile the facts around it point the other way. A person familiar with the partnership says it is still in place. A Replit blog post from late September described Claude models running in the production environment of Replit Agent, and a Replit product employee posted about using Opus around the same time. Replit announced its OpenAI partnership only in August, and nothing publicly suggested a departure from Anthropic before this talk.

Why the headline overreaches

"No longer working with" can mean a full contract end, a reduced share of traffic, or a negotiating position. A CEO on stage has every reason to praise a new partner. Until Replit clarifies, the accurate reading is this: the company is shifting toward OpenAI and says so loudly, while the scale of the Anthropic exit is unproven. Treat it as a signal about vendor mix, not as a verdict on which model is better.