Replit is making OpenAI its preferred AI partner, but that doesn’t mean Anthropic is disappearing from its coding platform.
Speaking at TechCrunch Disrupt, Replit CEO Amjad Masad said the company was no longer working with Anthropic, citing cost, performance, and partnership considerations behind its shift toward OpenAI. He later clarified that Replit remains model-agnostic and that users can still access Anthropic models.
“I think ChatGPT is better now,” Masad said. “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.”
The announcement caused immediate ripples across the software industry, particularly after Masad noted onstage that Replit had recently been one of Anthropic’s top five customers. However, both company insiders and Masad quickly stepped in to clarify the scope of the breakup.
The decision highlights how quickly AI coding platforms can reconsider their preferred providers as model capabilities and costs change. For developers and enterprise customers, it also raises questions about how those choices affect the tools they rely on.
Clarifying the default setting
While Masad’s onstage comments suggested a total separation, insiders familiar with the matter told TechCrunch that the partnership is still active. Masad later took to X to clarify that Replit remains open to multiple providers.
“Replit remains model agnostic,” Masad clarified on X. “Our users can choose Anthropic models as well as many other supported models… OpenAI is now our preferred partner powering Agent based on performance, cost, and strength of our partnership.”
The distinction is important. Replit is prioritizing OpenAI as the preferred provider for its Agent product, but users can still choose Anthropic and other supported models.
That gives developers continued access to multiple AI providers even as Replit changes which company it favors for its coding agent.
Why Replit’s decision matters for AI model providers
Replit’s rapid pivot illustrates an uncomfortable reality for foundation model labs: developer loyalty belongs to runtime economics, not brand affinity. Because coding agents consume enormous volumes of tokens through recursive loops, multi-file orchestration, and testing, developer platforms are exceptionally sensitive to inference margins.
By shifting the default engine, Replit demonstrates how application-layer platforms can commoditize model providers. Even without breaking contracts or ditching API access, an interface layer can instantly redirect tens of millions of queries with a single configuration tweak.
Foundation labs that lack proprietary distribution risk becoming interchangeable back-end utilities that compete ruthlessly on pricing concessions and unit economics.
Enterprise ambitions and the agentic habitat
The timing coincides with Replit’s aggressive push upmarket. Masad disclosed that enterprise contracts now account for roughly 15% of Replit’s revenue, up from 2% last year, with B2B clients driving more than half of overall sales. Backstage, Masad said he expects enterprise and government deals to contribute over half of the company’s revenue in 2027.
Scaling that business requires defensive engineering. Masad emphasized that autonomous systems require robust environments to operate securely.
“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,” Masad said, per TechCrunch.
More must-read AI coverage
- SS&C Intralinks DealCentre AI vs. Datasite: Which platform is built for the future of dealmaking?
- SS&C Intralinks FundCentre AI vs. Juniper Square: Which platform better supports modern private markets fund managers?
- Why Data, Not Models, Determines AI Success
- The Rise of the AI-Native Factory: How Physical AI Is Transforming Manufacturing
What this means for developers and engineering teams
For software creators and enterprise teams building on Replit Agent, the shift toward OpenAI could have practical implications for how they develop, test, and manage AI-generated code.
Builders relying on autonomous workflows may notice subtle differences in generated code syntax, logic structuring, and edge-case handling as OpenAI’s models take over prompt execution. Even when models perform similar functions, their approaches to debugging, interpreting instructions, and generating solutions can vary, potentially affecting the consistency of existing development workflows.
The change also raises questions about AI governance. Teams with internal security policies or governance frameworks approved specifically for Anthropic’s models should review their Replit configurations to understand which provider processes their requests.
This is particularly important for organizations handling proprietary code, sensitive business information, or projects subject to strict data-handling requirements.
Engineering leaders should proactively verify whether their workspaces rely on the automatic OpenAI default or manually lock their configurations to Claude models to avoid compliance friction. They should also consider testing critical workflows when switching providers to identify potential differences in code quality, reliability, and performance.
Ultimately, Replit’s decision highlights why model flexibility matters for enterprise development teams. Having access to multiple providers offers more choice, but organizations still need visibility into which models their tools use and how changes could affect security, compliance, and day-to-day development.
Other news: Nvidia is expanding its commitment to open AI infrastructure, contributing Vera Rubin rack designs and open-source AI factory software to the Open Compute Project.