Replit’s Bold AI Shift: OpenAI Dethrones Anthropic as Default

Key Takeaways

• Strategic AI Shift:Replit, a prominent AI coding platform, has officially transitioned its primary AI model partnership from Anthropic to OpenAI, citing superior performance, cost-efficiency, and a stronger partnership.
• Anthropic’s Reduced Role:While Replit remains “model agnostic,” allowing users to access Anthropic models, OpenAI’s ChatGPT is now the default for core functionalities like Replit Agent, a move expected to significantly reduce Replit’s spend with Anthropic.
• Growth Amidst Controversy:The partnership shift, which has been met with skepticism by some insiders, occurs as Replit reports substantial growth in its enterprise revenue, underscoring its evolving market strategy and commitment to AI safety.

Replit Pivots AI Allegiance, Elevating OpenAI Over Anthropic Amidst Growth and Debate

The dynamic landscape of AI partnerships saw a significant shake-up as Replit, the popular cloud-based AI coding platform, announced a strategic shift in its core AI model provider. Amjad Masad, Replit’s CEO and co-founder, revealed at TechCrunch Disrupt on Thursday that the company is no longer working with Anthropic as its primary partner, instead favoring OpenAI for its foundational AI capabilities.

This pivot marks a notable change for Replit, a platform that empowers startup founders and developers to build and deploy applications directly from their browsers. Masad articulated the decision as multi-faceted, driven by a confluence of factors including operational costs, the strength of the partnership, and crucially, performance metrics. “I think ChatGPT is better now,” Masad stated, highlighting that Replit’s internal benchmarks, alongside public assessments like Sol and Astra, consistently show OpenAI’s models performing superiorly for “building vibe-coded applications from scratch.”

The Strategic Rationale: Performance, Cost, and Partnership

Replit’s decision to shift its primary AI allegiance from Anthropic, a company it previously considered a top-five customer, to OpenAI underscores the intense competition and rapid evolution within the large language model (LLM) ecosystem. For a platform like Replit, which deeply integrates AI into its coding environment, the choice of underlying models directly impacts user experience, development efficiency, and ultimately, its own operational overhead.

Masad’s emphasis on “cost, partnership, and performance” as the driving forces behind the change provides a clear window into the strategic considerations of AI-native companies. In the highly competitive and often costly world of AI, even marginal improvements in model efficiency or reductions in API call expenses can translate into significant savings at scale. Furthermore, the notion of a “better partner” suggests that the collaboration with OpenAI may offer more favorable terms, deeper technical support, or a more aligned roadmap for future innovation compared to the previous arrangement with Anthropic.

Nuance in Disengagement: Default Over Exclusion

Despite Masad’s initial onstage comments suggesting a more definitive break, subsequent clarifications painted a picture of a more nuanced transition. Post-Disrupt, Masad confirmed that Anthropic hasn’t been entirely removed from Replit’s ecosystem. Customers retain the option to access Anthropic models on the platform; however, OpenAI’s ChatGPT is now set as the default. This seemingly minor adjustment holds significant weight.

In software, defaults often dictate the path of least resistance, guiding the vast majority of users’ choices. By making OpenAI the default, Replit effectively steers its user base towards ChatGPT for most AI-powered coding tasks, thereby significantly reducing the volume of API calls—and associated expenditure—with Anthropic. This strategic move allows Replit to maintain its “model agnostic” posture, offering choice, while pragmatically optimizing for its preferred partner and performance benefits.

Masad reiterated this position to TechCrunch, stating, “Replit remains model agnostic. Our users can choose Anthropic’s Claude or other supported models…OpenAI is now our preferred partner for powering Replit Agent, based on performance, economics, and the strength of our partnership.” This clarifies that while flexibility is offered, the strategic preference for OpenAI for core functionalities like Replit Agent—the platform’s AI coding assistant—is unequivocal.

Contradictions and the Competitive Undercurrent

The announcement, however, was not without its points of contention. Insiders familiar with the Anthropic-Replit partnership reportedly rejected Masad’s onstage comments, suggesting the partnership was still active. Compounding this, evidence on Replit’s own digital channels indicated recent promotion of Anthropic products. As recently as late September, a Replit blog post cited Anthropic’s Claude Fable 5 and 5.1 running in Replit Agent’s production environment, and a Replit product employee promoted Anthropic’s flagship Opus 5.5 model on X around the same time.

This discrepancy highlights the fast-moving nature of AI partnerships and potentially, the challenges in communicating internal strategic shifts externally. It also underscores the fierce competition among LLM providers to secure and retain high-profile platform integrations, with companies like Replit representing significant revenue streams and validation for model performance.

Replit’s Broader Trajectory: Enterprise Growth and AI Safety

Beyond the intricacies of its AI model partnerships, Masad also shed light on Replit’s impressive growth trajectory, particularly in the enterprise sector. While specific Annual Recurring Revenue (ARR) numbers were not disclosed, the shift in revenue composition is telling. “Enterprise was like 2% of revenue last year,” Masad noted, adding, “This year, I would say like 15% of our revenue is enterprise contracts, and more than 50% of our revenue is B2B.”

This surge in enterprise adoption signals Replit’s maturing market strategy, moving beyond individual developers and startups to attract larger organizations. Backstage at Disrupt, Masad projected an even more ambitious future, expecting over half of Replit’s revenue to originate from enterprise and government customers by 2027. This expansion into B2B and enterprise markets often brings greater revenue stability, larger contract values, and new demands for security and compliance, which influence platform development and partnership choices.

Furthermore, Masad touched upon the critical importance of AI safety, a topic increasingly paramount as AI systems become more powerful and pervasive. He articulated a vision where AI agents require a robust “habitat” to operate effectively and safely. “In the same way that us humans need a certain environment to work — you need a nice office, maybe nice good food in the office, air-conditioning, computer — agents need a habitat for them to exist and do good work,” he explained. This “habitat” encompasses good orchestration, error recovery, safety and security, sandboxing, networking, identity, and authorization. This philosophical stance reflects Replit’s commitment not just to enabling AI-powered coding, but to fostering a secure and responsible environment for AI development and deployment, a crucial factor for its growing enterprise clientele.

This article was originally published on October 16 at 2:22 pm PT. It has been updated to include more context around Replit’s partnerships with Anthropic and OpenAI.Its title has been updated, as well.

The Bottom Line

Replit’s strategic decision to prioritize OpenAI over Anthropic for its primary AI model partnership underscores the relentless competition and dynamic nature of the AI ecosystem. Driven by a pragmatic assessment of performance, cost, and partnership strength, this pivot, while contentious to some, positions Replit to optimize its core offerings for its rapidly growing user base and expanding enterprise market. As Replit continues its impressive growth trajectory and champions AI safety, its flexible yet decisive approach to AI model integration will likely serve as a blueprint for other platforms navigating the ever-evolving landscape of artificial intelligence. The message is clear: in the race to build the future of coding with AI, agility in partnership and a relentless focus on optimization are paramount.