What happened
Innovation Village reports that Replit has replaced Anthropic as the default model provider for its AI coding agent, following an August partnership with OpenAI. CEO Amjad Masad cited model performance, operating costs and commercial partnership strength. He later clarified that Replit remains model agnostic: customers can still choose Claude and other supported models. The report also says enterprise customers' share of Replit revenue rose from about 2% in 2025 to approximately 15% in 2026, though Replit has not disclosed annual recurring revenue.
Innovation Village reports that Replit selected OpenAI as its preferred partner for powering Replit Agent, replacing Anthropic as the platform's default provider. Masad attributed the decision to performance, cost and the broader commercial partnership. He said OpenAI models currently perform better for building applications from scratch, including natural-language "vibe coding" workflows.
The change does not remove Anthropic models from Replit. According to Innovation Village, Masad later clarified that customers can still select Claude and other supported models, and described Replit as model agnostic. The report also cites references to Anthropic models in Replit's production environment in late September, but says those references do not establish the current extent of the commercial relationship.
Innovation Village reports that Replit evaluates models with an internal and publishes a public benchmark, but the source does not provide benchmark scores or independently verified comparative results. It also reports that enterprise contracts represented about 2% of revenue in 2025 and approximately 15% in 2026; Replit's total revenue and ARR remain undisclosed.
Source details: innovation-village.com ↗
Why it matters
Making OpenAI the default can redirect a substantial share of Replit Agent usage and spending while changing how users experience AI-assisted software development. The decision also illustrates how coding platforms weigh model quality, costs, reliability and commercial terms rather than relying on results alone. Replit's reported enterprise growth could increase demand for stronger sandboxing, identity, authorization and administrative controls as agents write and execute code in organizational environments. Innovation Village's account is a secondary report, and AI Understanding has not independently verified Replit's commercial terms, benchmark results, revenue figures or the current scope of its Anthropic relationship.
A default provider influences user behavior even when alternatives remain available. Users who do not manually change the setting may generate more workloads through OpenAI, affecting model-provider economics and Replit's operating costs.
The decision is consequential for AI coding products because model output is only one part of the system. Agents that execute code or access files and services also require isolation, networking controls, identity management, authorization and failure recovery. The source discusses these needs but provides no independent assessment of Replit's safeguards.
The reported shift also reflects competition among model providers to become infrastructure partners for application-building platforms. The source does not document the partnership's contract terms, exclusivity provisions, pricing, service-level commitments or any independently measured improvement in Replit Agent.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
crm_get_transaction(id='4092').An agent must create a draft calendar event for Tuesday at 2 p.m. Which evidence would establish the requested result?
What to watch next
The important questions are whether OpenAI remains the default over time, how Replit's models perform on real application-building tasks, and whether pricing or reliability changes alter the platform's economics. Watch for documentation showing which models are available to which customers, whether Claude access has limitations, and whether enterprise and government growth leads to new security, governance or deployment features. Replit's stated expectation that enterprise and government customers will exceed half of revenue in 2027 is a forecast, not an established result.
Replit's public model-selection documentation and results may clarify whether OpenAI's preferred status affects availability, , quality or price for different users.
The extent to which Claude and other models remain accessible is not fully specified. It is also unknown whether all plans, enterprise customers and workflows receive the same choices.
Replit's enterprise strategy will likely make security controls and predictable costs more important. The claim that enterprise and government customers will exceed half of revenue in 2027 remains an expectation rather than a verified outcome.