Replit Says It Dropped Anthropic for OpenAI, But Details Are Fuzzy

Replit’s CEO says the company is done with Anthropic. That statement, made onstage at TechCrunch Disrupt 2026, has created confusion. The company’s own blog posts and an unnamed insider tell a different story.

Here’s what we know. On Oct 16, 2026, TechCrunch reported that Replit CEO and co-founder Amjad Masad said the company is “no longer working with Anthropic.” Masad gave three reasons: cost, partnership, and performance. His words: “It was a decision based on both cost, you know, partnership, and performance.”

That sounds clear. But it’s not that simple.

The Timeline of a Quiet Switch

Replit is a browser-based app-building platform. For a long time, it used Anthropic’s Claude models to power its core features. That was public knowledge. Then, in August 2026, Replit announced a new partnership with OpenAI. That announcement did not mention Anthropic at all.

The problem? Replit never publicly said it had ended its Anthropic relationship. One report describes it as a “quietly transitioned” arrangement. No press release. No blog post. No farewell to Claude.

That silence is now causing problems. Because the CEO’s statement at Disrupt contradicts what the company itself published just weeks earlier.

The Pushback: A September Blog Post

Let’s look at the evidence. In September 2026, Replit published a blog post. That post explicitly cited Anthropic’s Claude Fable 5 and Claude Fable models. That means Replit’s own products were using Anthropic technology as recently as September.

So we have a conflict. The CEO says “no longer working with Anthropic” in October. The company’s own blog post from September says it’s using Anthropic models. That’s a short window. But it’s a real contradiction.

Then there’s the unnamed insider. This person, described as “familiar with the partnership,” said the Anthropic relationship was still in place. They pushed back on Masad’s claim. No name, no title, no official statement. But the source exists.

So what do we believe? The CEO’s onstage remark? The blog post? The anonymous insider?

Why This Matters: Cost, Performance, and Partnerships

Masad’s stated reasons are worth unpacking. Cost is a big one. AI model usage is expensive. Replit is a platform that lets users build apps. If it’s serving millions of requests, model costs can explode. Switching to a cheaper provider makes business sense.

Performance is another factor. Different models have different strengths. Maybe Claude was slow for certain tasks. Maybe OpenAI’s models were faster or more accurate. Masad didn’t give specifics. He just said “performance” was a factor.

Partnership is the third reason. That’s vaguer. It could mean strategic alignment. It could mean OpenAI offered better terms. It could mean Anthropic wanted something Replit didn’t want to give. Without more detail, we’re guessing.

The Dispute: Who’s Right?

Here’s where it gets tricky. The insider says the partnership is still in place. The blog post shows Anthropic models in use. The CEO says it’s over.

Possible explanations:

1. The CEO misspoke. He meant the partnership was ending, not that it had already ended.
2. The insider is wrong or has outdated information.
3. The blog post was written before the switch but published later. That happens.
4. Replit is using both providers right now, with a plan to phase out Anthropic.
5. The CEO’s statement was aspirational, not factual.

We don’t have enough information to pick one. This is a disputed claim. Both sides have evidence. The only hard facts are: the CEO said it’s over, the blog post says it’s not, and an insider says it’s not.

What Replit and Anthropic Say (and Don’t Say)

Neither company has issued an official statement. Anthropic has not commented publicly. Replit has not clarified its CEO’s remarks. The silence is notable.

In tech, partnerships end all the time. Companies switch AI providers regularly. But usually, they announce it. They say “we’re moving to X.” They explain why. They thank the old partner.

Replit did none of that. It announced OpenAI in August. It quietly kept using Anthropic in September. Then the CEO said the Anthropic relationship was over in October.

That pattern suggests either a messy transition or a miscommunication. Neither is a good look.

The Business Context: Why This Deal Matters

Replit is a developer tool. It competes with other app-building platforms. Its AI models are a core feature. If users type a prompt and get a bad result, they leave. So model choice is critical.

OpenAI is the market leader. Anthropic is a strong challenger. Both have loyal fans. Some developers prefer Claude’s style. Others like OpenAI’s speed. Replit’s choice affects its user experience.

If Replit truly dropped Anthropic, that’s a win for OpenAI. It’s a loss for Anthropic. If Replit didn’t drop Anthropic, then the CEO’s statement is a serious error.

Either way, this is a story about transparency. Replit needs to clarify its position. Users deserve to know what models power their apps. Investors need to understand the company’s strategy.

What to Watch Next

Watch for an official Replit statement. If the company confirms the switch, look for a timeline. When did it happen? What models are they using now? If the company denies the switch, then Masad has some explaining to do.

Also watch Anthropic’s response. A quiet partner doesn’t usually stay quiet. They may issue a statement or update their customer list.

Finally, watch Replit’s blog. If the next post mentions OpenAI models only, that’s a sign. If it still mentions Claude, then the CEO’s claim is shaky.

This story is not resolved. It’s a he-said, she-said with a blog post as a third party. Until Replit or Anthropic speaks clearly, treat the CEO’s statement as a claim, not a fact.