Replit makes OpenAI its default over Anthropic, CEO says: AI Coding

Replit, the popular AI coding platform, has made OpenAI its default artificial intelligence model, sidelining its previous primary partner, Anthropic. CEO and co-founder Amjad Masad announced the significant shift during TechCrunch Disrupt, attributing the decision to OpenAI's superior performance, cost efficiency, and a stronger partnership. While users can still access Anthropic models, the default change is poised to substantially redirect Replit's AI spending.

Masad initially stated onstage that Replit was "no longer working with" Anthropic. However, he later clarified to TechCrunch that Anthropic models remain available on the platform, but OpenAI's ChatGPT would now serve as the preferred default for powering Replit Agent. This move is expected to significantly reduce Replit's expenditure with Anthropic, as a majority of users typically adhere to the default options.

Strategic Pivot from a Key Partner

Prior to this strategic pivot, Replit heavily relied on Anthropic's Claude models to operate its platform, with the partnership reportedly fueling substantial revenue growth for the company. Masad even noted Replit had recently been among Anthropic's top five customers. The company had announced a partnership with OpenAI in August, though without indicating an end to its collaboration with Anthropic.

Despite the CEO's on-stage remarks, sources familiar with the Anthropic-Replit collaboration disputed the claim of the partnership being entirely severed, insisting it was still active. Recent evidence, including Replit blog posts and social media updates from late September, showcased the promotion and use of Anthropic's Claude models within Replit's production environment.

Performance, Cost, and Partnership Drive Decision

Masad emphasized that the decision was multifaceted, driven by internal benchmarks indicating OpenAI's "Sol and Astra" models consistently outperforming Anthropic's offerings in building "vibe-coded applications from scratch." He asserted, "I think ChatGPT is better now." Replit also publishes a public benchmark that supports these findings.

Beyond performance, the economics of the partnership and the overall strength of their relationship with OpenAI were critical factors. Masad stated, "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 maintains that it remains "model agnostic," assuring users they retain the flexibility to select Anthropic's Claude or other supported models on the platform.

Replit's Expanding Enterprise Footprint

Beyond the AI partnership news, Masad also shared updates on Replit's impressive business expansion. The company is experiencing a significant surge in enterprise revenue, which constituted just 2% of total revenue last year but has climbed to approximately 15% this year. Furthermore, B2B contracts now account for over 50% of Replit's overall revenue.

Masad projected that by 2027, more than half of Replit's revenue would originate from enterprise and government clientele, signaling a strategic shift towards larger organizational clients and sustained growth in new markets.

Emphasizing AI Safety and Agent Habitats

During his appearance, Masad also touched upon the critical importance of robust AI safety measures. He likened the ideal environment for AI agents to human workspaces, requiring "a habitat for them to exist and do good work." This habitat, according to Masad, necessitates effective orchestration, error recovery, stringent safety and security protocols, sandboxing, networking, identity management, and authorization, all essential for the safe deployment of AI.

FAQ

Q: Has Replit completely stopped working with Anthropic?

A: No. While Replit has made OpenAI its default AI model, users can still access Anthropic's Claude models and other supported options on the platform. The shift primarily designates OpenAI as the preferred partner for powering Replit Agent.

Q: Why did Replit make OpenAI its default over Anthropic?

A: Replit CEO Amjad Masad cited several reasons, including OpenAI's superior performance based on internal and public benchmarks, better cost efficiency, and a stronger overall partnership with OpenAI.

Q: What does this change mean for Replit users?

A: For most users, OpenAI's models will now be the pre-selected default when using Replit Agent. However, users still have the flexibility to manually choose Anthropic's Claude or other available models according to their preference.