Replit's CEO says it left Anthropic, but an informed source disputes the split

A provider exit claim with an immediate contradiction

Replit co-founder and CEO Amjad Masad says the coding platform is no longer working with Anthropic, according to TechCrunch’s October 16 report from Disrupt. But the same report cites an unnamed person familiar with the partnership who says it is still in place. The commercial split therefore remains disputed.

TechCrunch reports that Masad made the remarks Thursday and attributed the decision to cost, partnership and performance. It quotes him saying, “I think ChatGPT is better now,” and saying Sol and Astra consistently perform better at building vibe-coded applications from scratch in the company’s assessments.

Those are Masad’s explanations and performance claims. They do not establish that OpenAI models outperform Claude for every coding workload, or that a particular customer’s Replit session has already switched providers.

September production evidence complicates the picture

Replit’s own September 29 engineering post documents Anthropic models in its production agent: its table covers Fable 5 in August and Fable 5.1 alongside GPT-6 Astra in September. The post describes different delegation behavior by model and argues for a harness that lets each model decide when to hand work to subagents.

That dated evidence shows recent Anthropic use. It cannot, by itself, settle whether a commercial relationship subsequently ended or distinguish termination from a change in preferred models. TechCrunch says Replit announced an OpenAI partnership in August, but that announcement had not indicated it was leaving Anthropic.

The publication had asked Replit for clarification and said it would update its story if it received a response. Until that conflict is resolved, the supported news is a CEO’s claimed exit and a source’s explicit disagreement, rather than a confirmed Anthropic contract cancellation or a complete removal of Claude from Replit.