Replit Adopts ChatGPT as Main Model Despite Anthropic

Key Takeaways

1Replit makes OpenAI's ChatGPT its default model

2The partnership with Anthropic is controversial, but access to its models remains possible

3The share of revenue from B2B and the public sector is growing significantly

Why it matters : These developments reflect a strategic repositioning of Replit regarding its AI partners and target markets.

The CEO of Replit claims to no longer work with Anthropic and has made ChatGPT the default model on the platform. However, access to Anthropic's models remains possible, and sources suggest that the partnership is not broken. Replit also reports a rapid increase in B2B revenue and aims for a majority of public and government clients by 2027.

The status of the partnership with Anthropic is contested

Individuals close to the partnership between Anthropic and Replit dispute Amjad Masad's statements regarding the end of the collaboration. An anonymous source claims that the partnership is still in place. Recent elements from Replit's website, blog posts, and social media publications continue to highlight Anthropic's products. Claude Fable 5 and Claude Fable 5.1 are used in the production environment of Replit Agent. Around the same time, a product employee from Replit posted on X about using Anthropic's Opus 5.5 model and praised its capabilities.

What’s changing on Replit: ChatGPT by default, access to Claude maintained

Following TechCrunch Disrupt, Amjad Masad indicates that users still have the option to use Anthropic's models on the platform, but OpenAI's ChatGPT is now selected as the default model. He also clarifies that Replit maintains a neutral approach regarding models and that users can opt for Claude from Anthropic or other compatible models. OpenAI is presented as the preferred partner to power Replit Agent, due to performance, cost-effectiveness, and the strength of the partnership.

Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

Reasons cited: costs, benchmarks, and performance

Amjad Masad explains that the decision to favor OpenAI is based on several criteria, including cost, performance, and the quality of the partnership. He believes that ChatGPT is currently superior. Replit uses an internal benchmark to evaluate metrics such as performance and also publishes a public benchmark. According to Masad, the Sol and Astra models achieve better results for creating applications coded from scratch. These factors influenced the choice of partner.

Economic consequences and history of models at Replit

The shift to ChatGPT as the default model could reduce Replit's spending with Anthropic, as most clients will retain the default option. Before the agreement with OpenAI, Replit used Anthropic's Claude models to operate its platform. This partnership is said to have contributed to significant revenue growth, and Replit was among the top five clients of Anthropic until recently, according to Masad. On the commercial front, Masad indicates that Replit is not yet disclosing ARR figures but notes that the share of revenue from the enterprise sector has increased from about 2% last year to around 15% this year for enterprise contracts, and more than 50% of revenue now comes from B2B. Masad anticipates that by 2027, more than half of the revenue will come from public and government sectors.

Security framework for agents described by Replit

Amjad Masad emphasizes the importance of robust security measures for AI deployment. He describes that AI agents require a suitable environment that includes orchestration, error recovery, security, sandboxing, as well as a network, identity management, and authorization to ensure safe implementation.