Replit CEO says the company is no longer working with Anthropic

In short: Replit’s CEO says the company has stopped working with Anthropic and is now leaning on OpenAI instead.

What happened

Replit is a software tool that lets people build and run apps in a web browser, kind of like writing and publishing a document without installing anything. It also includes an AI helper that can write and edit code.

On stage at TechCrunch Disrupt, Replit CEO and co-founder Amjad Masad said Replit is no longer working with Anthropic, the AI company behind the Claude models. Masad said the change was based on several factors, including cost, partnership terms, and performance.

Masad also said “ChatGPT is better now,” and described Replit’s use of benchmarks, which are basically scorecards used to compare how well different AI models do on the same tasks. He said Replit’s own tools, including “Sol” and “Astra,” have performed better when building “vibe-coded” apps from scratch, meaning apps created with a lot of natural language instructions rather than step by step programming.

Replit announced a partnership with OpenAI in August. Before Masad’s comments, there was no clear public sign that Replit had ended its relationship with Anthropic.

There is some conflicting information. TechCrunch reported that one person familiar with the partnership said it was still in place. TechCrunch also pointed to a September Replit blog post that mentioned Anthropic models running in Replit Agent, which suggests Anthropic technology may have been used recently.

Why it matters

For people using Replit, the AI “brain” behind the coding helper can affect price, speed, and the quality of suggestions. This also shows how quickly AI partnerships can shift as companies compare results and costs.

Source: TechCrunch AI