Replit CEO Says Company Ditched Anthropic for OpenAI — But Insiders Push Back

Key takeaway: Replit CEO Amjad Masad said at TechCrunch Disrupt (Oct 16, 2026) that the coding platform no longer works with Anthropic, citing cost, partnership, and performance — with OpenAI’s models beating Claude on Replit’s internal benchmarks. People familiar with the partnership dispute the claim, though: Replit was still using Claude models in September. The full story is still unfolding.

What Masad said on stage at TechCrunch Disrupt

Speaking onstage at TechCrunch Disrupt on Thursday, October 16, Replit co-founder and CEO Amjad Masad was asked whether the OpenAI partnership Replit announced in August meant the company had stopped working with Anthropic. “Yeah. I mean, yeah,” Masad answered, per TechCrunch.

Masad said the decision came down to “a number of different reasons, including cost.” He added that OpenAI’s models now perform better for Replit’s workloads: “I think ChatGPT is better now.” Replit runs an internal benchmark and also publishes a public one, and Masad said two models — which he named as Sol and Astra — “consistently perform better on building vibe-coded applications from scratch” than Anthropic’s offerings.

Summing up the move, Masad said: “It was a decision based on both cost, partnership, and performance. And for all these reasons, we just felt like they were a better partner for now.”

Replit sells a cloud-based software development platform where users can build and deploy apps directly from their browsers. Under its earlier partnership with Anthropic, the company used Claude models to power the platform — a collaboration reported to have helped Replit grow its revenue significantly. When Replit announced the OpenAI partnership in August, there was no indication it was ending its work with Anthropic.

The dispute: insiders say the partnership is still in place

Masad’s account was immediately challenged. One person familiar with the Anthropic–Replit partnership, who asked not to be named, told TechCrunch the partnership was still in place.

There is also public evidence that Replit was using Anthropic products as recently as a few weeks ago. A Replit blog post from late September listed 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 own product employees posted on X about using Anthropic’s flagship model Opus 5.5 and promoting its capabilities.

TechCrunch says it has reached out to Replit for clarification and will update its reporting if it hears back. As of publication, neither company has issued an official statement confirming or denying the split.

Why this matters

For developers who build on Replit, the practical impact is the model underneath the hood. Replit’s agent writes, tests, and deploys code on a user’s behalf; which lab’s models do that work shapes output quality, speed, and the pricing users ultimately see. If OpenAI really has displaced Claude across Replit’s platform, it marks one of the highest-profile model-partner switches in the AI coding space this year.

The economics matter too. For developer platforms, API costs on millions of coding requests can make or break unit economics — Masad naming cost first, ahead of partnership and performance, is a signal about where margins live in AI coding tools. Microsoft’s custom AI chips for Copilot show the same pressure from the hardware side: everyone is hunting cheaper inference.

It is also a rough week for Anthropic’s reputation with large customers. Masad added onstage that Replit was one of Anthropic’s top five customers — losing a marquee account would sting. That makes the insider pushback significant: the difference between “Replit quietly migrated away months ago” and “the partnership is still in place” is the difference between a done deal and a live negotiation playing out in public.

Analysis: what to watch

Analysis (not yet verified): The two accounts here are not necessarily mutually exclusive. Replit could have stopped paying for or prioritizing Anthropic models while some contractual partnership technically remains — a nuance that would let both Masad and the insider be partially right. Watch for three things: an official Replit statement, whether Anthropic models disappear from Replit’s agent settings, and whether Anthropic responds at all. Until then, treat the split as claimed but unconfirmed. OpenAI, meanwhile, keeps collecting platform wins — adding to a year in which it is already the center of industry-shaking disputes, from state lawsuits over ChatGPT safety to enterprise deals that reshape its rivals’ economics.

FAQ

Did Replit really stop working with Anthropic?

CEO Amjad Masad says yes — he told TechCrunch Disrupt on October 16, 2026 that Replit no longer works with Anthropic, citing cost, partnership, and performance. However, a person familiar with the partnership says it is still in place, and Replit’s own September blog listed Claude models in its production environment. Neither company has issued an official statement, so treat the split as claimed but unconfirmed.

Why did Replit choose OpenAI over Anthropic?

According to Masad, the decision came down to cost, partnership, and performance. He said Replit’s internal and public benchmarks show OpenAI’s Sol and Astra models consistently perform better at building vibe-coded applications from scratch than Anthropic’s offerings. Replit announced its OpenAI partnership in August 2026 without mentioning any change to the Anthropic relationship.

What Anthropic models did Replit use?

Under its partnership with Anthropic, Replit used Claude models to power its coding platform. As recently as late September 2026, a Replit blog post cited Claude Fable 5 and Claude Fable 5.1 as running in Replit Agent’s production environment, and a Replit product employee posted about using Anthropic’s Opus 5.5 model.

How does the Replit–Anthropic split affect Replit users?

In the short term, probably not much — your projects and code are unaffected by which lab’s models run underneath. Over time, the change could shift the quality and speed of Replit’s AI-generated code, and API economics may influence pricing. If you notice different model names in Replit’s agent settings, that is the switch becoming visible.

What happens next?

TechCrunch has asked Replit for clarification and says it will update its story with any response. Watch for an official statement from Replit or Anthropic, and for changes in Replit’s agent model lineup. Until one company speaks on the record, the situation remains unresolved.

Sources

• Replit is no longer working with Anthropic, CEO says — TechCrunch, Oct 16, 2026
• Replit CEO Masad Says Company No Longer Works With Anthropic — Unite.AI, Oct 16, 2026