Replit has made OpenAI’s ChatGPT the default model for Replit Agent, citing performance, cost, and the strength of the partnership. Customers can still choose Anthropic’s Claude or other supported models.
TL;DR: Replit set ChatGPT as the default for Replit Agent because it considers it better in terms of performance, cost, and partnership, but Anthropic has not been removed. Users can still choose Claude themselves. Put simply, the default model is like the “front-of-store option” that most people tend to pick first.
However, the status of Replit’s partnership with Anthropic remains unclear. While the CEO said they had stopped working together, insiders say a partnership still exists. Replit’s website and posts have also continued promoting Anthropic’s models until recently.
How Replit Changed Its Default and Why It Matters
Replit has set OpenAI ChatGPT as the default model for Replit Agent, citing performance, economics, and partnership considerations. Users can still choose Anthropic’s Claude or other supported models.
This change matters because most customers tend to use the model selected by default, which could reduce Anthropic’s usage and Replit’s costs, even though Anthropic has not been removed from the platform used to develop and deploy apps through a browser.
When Coding Tools Choose the Model for Users
People who open Replit and start building an app may use the default model without selecting one themselves. With ChatGPT becoming the default, most users are therefore more likely to start with OpenAI’s model.
Anthropic remains available as an option, but the default position affects the actual experience and usage because Replit says ChatGPT is currently the primary partner for Replit Agent.
If you ask me, I see this as more than simply switching models. It sets the direction of user behavior from the very first step. The default model is like the older sibling given the main stage, while Claude remains in the system but acts like the younger sibling whom users must seek out themselves.
Where Replit Is Positioning OpenAI in Its Business Direction
Replit is a cloud-based software development platform that lets users build and deploy apps from within a browser. This makes AI models (artificial intelligence systems that help create and edit code) an important part of the user experience, with OpenAI becoming the primary partner Replit uses for Replit Agent.
However, Replit still says that the company is not tied to any single model. Users can still choose Anthropic’s Claude or other supported models. The current direction is therefore to position OpenAI as the primary option while still allowing users to select models based on their needs.
From Claude to ChatGPT: What Changed at the Default Level
Replit has shifted from relying on Anthropic’s Claude to using OpenAI’s ChatGPT as the default for Replit Agent, citing performance, cost, and the strength of the partnership. Users can still choose Claude or other supported models.
| Factor | Anthropic's Claude | OpenAI's ChatGPT |
|---|---|---|
| Primary status | The model Replit relied on before its agreement with OpenAI | The current default model |
| Performance | Previously used to power the platform | Replit says it performs better at building apps from scratch |
| Cost and partnership | One of the reasons Replit reconsidered its usage | Selected based on cost and partnership |
| User access | Still available to choose | Set as the default |
How the Default Model Affects Real-World Usage
When building an app from scratch or having Replit Agent work in a vibe-coded manner (asking AI to create an app from a general description), the default model is ChatGPT because Replit says Sol and Astra performed better in this area’s benchmark (a comparative test).
People who want to use Claude can still select Anthropic themselves because Replit continues to support Anthropic’s models and emphasizes that it remains model agnostic. However, most users tend to use the model set as the default.
For enterprise customers, having ChatGPT as the default helps teams start working in the same direction. At the same time, organizations that require a specific model still have options, as Replit says the decision considered performance, economics, and partnership.
How Replit, Anthropic, and OpenAI Differ
All three companies play different roles within the same platform. Replit is the platform for developing and deploying apps through a browser, while Anthropic and OpenAI provide models supported by Replit.
| Factor | Replit | Anthropic | OpenAI |
|---|---|---|---|
| Role | Platform for developing and deploying apps | Provider of Claude models | Replit's model partner |
| Relationship with Replit | Platform developer | Former partner whose models remain accessible | New primary partner |
| Current status | Still a model-agnostic platform | Still available for users to choose | ChatGPT is the default for Replit Agent |
In summary, Replit has not completely removed Anthropic. It has chosen OpenAI as the default, while users can still switch to Claude if they prefer.
Strengths and Limitations of This Change
Pros
- +Uses internal benchmarks to measure performance before selecting a model
- +Considers cost, partnership, and performance
- +Users can still choose Claude
Cons
- −Information about the partnership with Anthropic remains contradictory
- −Replit has not disclosed its ARR figures (annual recurring revenue)
Costs Not Included in the Usage Price
Replit’s costs include more than the fees paid by users. They also include the cost of the AI models the company must cover. Choosing OpenAI as the default could therefore reduce expenses with Anthropic because most users tend to use the model selected by the system.
Another point to watch is the uncertainty surrounding the partnership. Although Replit says OpenAI is its new primary partner, the status of its partnership with Anthropic remains unclear.