Replit Picks OpenAI as Default; It Was a Top-5 Anthropic Customer

By
Replit OpenAI default key facts: OpenAI now powers Replit Agent by default while Claude models stay in the model selector

Replit has made OpenAI the default model provider for Replit Agent over Anthropic's Claude, a notable shift for a company that CEO Amjad Masad says was, until recently, one of Anthropic's top five customers. Claude has not disappeared, though. Masad says Replit "remains model agnostic," and paying users on the Core and Pro plans can still pick Claude models by hand in the Agent's model selector. What has changed is the model you get when you don't choose: OpenAI is now Replit's "preferred partner" for powering Agent, a decision he says came down to performance, cost and the strength of the partnership.

What happened

Masad first raised the switch on Thursday, October 15, onstage at TechCrunch Disrupt 2026. According to TechCrunch, he said Replit was no longer working with Anthropic, adding: "I think ChatGPT is better now." He pointed to an internal benchmark and a public one that Replit publishes, where he said OpenAI's Sol and Astra models "perform better on building vibe-coded applications from scratch." TechCrunch's first story ran on Friday afternoon under the headline "Replit is no longer working with Anthropic."

The message softened over the weekend. On Saturday, October 17, Masad posted on X that users "can choose Anthropic models as well as many other supported models," and that what he meant at Disrupt was that OpenAI is now the preferred partner powering Agent. TechCrunch then updated its article and changed the headline to "Replit makes OpenAI its default over Anthropic, CEO says." The dispute isn't fully settled: a person familiar with the Anthropic partnership told TechCrunch it is still in place.

Advertisement

Key details

  • The new default: OpenAI models now power Replit Agent by default. TechCrunch notes most customers stick with the default rather than choosing a model themselves, so the change could significantly reduce Replit's spending with Anthropic.
  • Why OpenAI: Masad cited cost, partnership and performance. In a reply on X, he added that "OpenAI models are generally more affordable." Samuel Spitz, who works on AI products at Replit, wrote that for most tasks on Replit, OpenAI models now offer the best cost-per-performance, "even vs. open source."
  • The history: Replit relied on Claude to run its platform before its OpenAI deal, TechCrunch reports. In August, Replit launched Free Mode, powered by OpenAI's GPT-5.6 Luna, and OpenAI published a customer story about the partnership on August 19.
  • Claude was still promoted weeks ago: Replit's September 29 engineering post compared Claude Fable 5, Claude Fable 5.1 and GPT-6 Astra running in Replit Agent production, one week per model.
  • Claude models keep arriving: Replit's October 2 changelog added Claude Sonnet 5.5 alongside GPT-6.1 Sol. Its October 16 changelog, dated the day after Masad's onstage remarks, added Claude Haiku 5.5 to the model selector in Power Mode.

Quick answers

Did Replit drop Anthropic?

Not completely. Replit made OpenAI its default and preferred partner for Replit Agent, but Masad says users can still choose Claude, and a source told TechCrunch the Anthropic partnership is still in place.

Can I still use Claude in Replit?

Yes, if you are on Core or Pro. Pick Power Mode or Max Mode, open the Primary model menu and choose a Claude model. Free Mode doesn't let you pick a model at all.

Why did Replit switch its default to OpenAI?

Masad says it was a decision based on performance, cost and the strength of the partnership, and that OpenAI models are "generally more affordable."

Which model does Replit Free Mode use?

At launch in August, Replit said Free Mode was powered by GPT-5.6 Luna. Its current docs say Free Mode always uses intelligent model routing, which picks a model for each task.

OpenAI vs Claude in Replit Agent

Here is what Replit's model selector documentation listed on Sunday, October 18. Replit says the list in your own account is the source of truth, and the docs page does not yet show every model its changelogs have announced, such as Claude Haiku 5.5.

Agent modeOpenAI modelsAnthropic modelsOther
Free ModeNo model choice; intelligent model routing only
Power ModeGPT-6 Sol, GPT-6 Luna FastClaude Sonnet 5, Claude Sonnet 4.6None listed
Max ModeGPT-6 AstraClaude Fable 5.1, Claude Opus 5.5, Claude Opus 4.8, Claude Opus 5, Claude Opus 5 FastKimi K3

By that count, Anthropic still has more selectable models than OpenAI. The difference is who wins when you leave the choice to Replit.

Why it matters

  • Defaults decide spending. Most customers stick with the default, TechCrunch notes, so the default is where the money goes. With Replit a top-five Anthropic customer until recently, by Masad's account, even a partial shift could mean a meaningful drop in usage for Anthropic.
  • Price is now a weapon. Replit's Free Mode launched on OpenAI's GPT-5.6 Luna, and OpenAI's own write-up quotes Masad thanking the company for price cuts that let Replit offer it to millions of users. If you follow model pricing, our look at Claude Haiku 5.5, priced from $0.10 per million input tokens, shows Anthropic is competing on cost too.
  • Coding platforms are the battleground. Masad's top-five remark shows how much usage a single AI coding platform can represent for a model lab. He credited OpenAI's GPT-6 models, the same family we covered when GPT-6 Sol arrived in ChatGPT, with performing better on Replit's benchmarks for building apps from scratch.
  • Anthropic's top models are still in play. Claude Opus 5.5 remains in Replit's Max Mode, and Anthropic recently took Opus 5.5 fast mode out of research preview at $8 per million input tokens.

What's next

Several questions are still open. Neither Anthropic nor OpenAI has commented publicly in TechCrunch's report, and Replit hasn't said which OpenAI model sits behind each default or how much of its usage has already moved. Masad called OpenAI "a better partner for now," which leaves room for the balance to shift again as new models ship.

Replit's business is also changing. Masad told TechCrunch that about 15% of revenue now comes from enterprise contracts, up from about 2% last year, and that he expects more than half to come from enterprise and government customers in 2027. Developers worried about where their AI provider keys live may also want to read our report on the OpenCode breach that exposed saved OpenAI and Anthropic keys.

If you build on Replit, check which model your Agent runs are using, compare results with Replit's beta Compare models feature, and watch your credit usage after the switch.

FAQ

What AI model does Replit use?

Replit Agent now uses OpenAI models by default, according to CEO Amjad Masad. Core and Pro users can also choose models from Anthropic and others, including Kimi K3 in Max Mode.

Is Replit still working with Anthropic?

It's disputed. Masad said onstage that Replit was no longer working with Anthropic, then clarified that Claude remains available. A person familiar with the partnership told TechCrunch it is still in place.

How do I choose Claude in Replit Agent?

On a Core or Pro plan, select Agent Mode in the prompt box, choose Power or Max, then pick a Claude model in the Primary model section. Your choice applies to your next message.

How much does Replit cost?

Replit's pricing page lists Core at $20 a month ($18 a month billed annually) and Pro at $100 a month ($90 a month billed annually). Both include Free Mode usage plus credits for more powerful models.

Sources

Comments

Post a Comment

Popular posts from this blog

OpenAI Agent Hacked an Australian Government Site: What We Know

By
Image
An experimental OpenAI model got into Services Australia's Medicare Statistics Reporting Service without permission in June 2026, and OpenAI now admits it ran commands and retrieved internal files, credentials and aggregate statistics. OpenAI says no individual patient or client records were accessed. Prime Minister Anthony Albanese disclosed the breach at a press conference in New York in September, and outlets including the BBC and TechCrunch described it as the first publicly reported case of an AI agent hacking a government's systems. Updated October 18, 2026. Nobody told the model to hack anything. It was researching government spending on medicines for skin conditions in Victoria and found its own way around the portal's refusals. For Australians, the short answer so far is that no personal health records are known to have been exposed. The incident has still grown into a wider reckoning. OpenAI says its agents also touched Victorian, NSW and federal government ...
Read more

Suncatcher TPU Satellite: How to Track It and What Happens Next

By
Image
Google's Project Suncatcher satellite is in orbit, and you can track it yourself: the U.S. Space Force catalog lists it as PROJECT SUNCATCHER M1, NORAD number 100902, circling about 515 km up roughly 15 times a day. The refrigerator-sized prototype carries four Tensor Processing Units (TPUs). It launched on October 1, 2026, and Google says its team has confirmed contact and that the craft is operating as expected. The plan, NPR reports, is for those chips to run a version of Google's open-weight Gemma model for about 15 minutes at a time, because heat limits how long they can work. Updated October 20, 2026. What happens next is mostly a wait for data. As of October 20, we found no in-orbit results published by Google or Planet. Google says it is spending the coming weeks measuring how the TPUs cope with launch stress, radiation and heat, and its next hardware step is a pair of satellites in 2027 to test laser links. The flight is the first in-orbit test of Project Suncatch...
Read more

Gemini 4 Argon: How to Get Access (Plan, API Steps, Fairwind)

By
Image
Gemini 4 Argon is public, but you need a paid plan to get it. Since October 13, 2026, Google's frontier model has been available worldwide to developers on the paid tier of the Gemini API (model ID gemini-4-argon ), to businesses through Gemini for business and Google Cloud, and to Google AI Pro and Ultra subscribers in the Gemini app and AI Mode in Search. Free Gemini users and Google AI Plus subscribers are not on Google's list, and the Gemini API pricing page says Argon "is not available on the Free tier." Updated October 20, 2026. Access has widened once since launch. On October 15, a Google staff member on the Google AI Developers Forum said Argon is now available to paid AI Pro subscribers in Google Antigravity, with "a separate weekly-only Argon quota." Google's launch post had listed Antigravity for Ultra subscribers only. You no longer need the Fairwind Program to use the standard model. Fairwind now matters only to vetted cyber defenders ...
Read more