Replit is no longer working with Anthropic, CEO says
Replit's CEO, Amjad Masad, announced at TechCrunch Disrupt that the AI coding platform has ceased its partnership with Anthropic due to cost, performance, and partnership reasons, favoring OpenAI's models.
Intelligence analysis by Gemini 2.5 Flash
Replit's CEO, Amjad Masad, publicly stated that the company has ended its collaboration with AI lab Anthropic, citing better performance and cost-effectiveness from OpenAI's models. This announcement, made at TechCrunch Disrupt, was contradicted by insiders who claim the partnership is still active, with recent evidence of Replit using Anthropic's products.
Replit, a company that helps people write computer code using AI, decided to stop working with one AI helper called Anthropic. Their boss said another AI helper, like a smarter friend named ChatGPT, was better and cheaper. But some people who know about the deal say Replit is still using Anthropic's tools, making it a bit confusing like when your parents say one thing and you see another.
Analysis
Replit's CEO, Amjad Masad, made a significant announcement at TechCrunch Disrupt, declaring that the AI coding platform has terminated its partnership with Anthropic. Masad attributed this decision to a combination of factors, including cost-effectiveness, the nature of the partnership, and the superior performance of alternative models, specifically mentioning OpenAI's ChatGPT. He indicated that Replit utilizes both internal and public benchmarks to assess AI model performance, and these evaluations consistently showed better results from OpenAI's offerings for building 'vibe-coded applications from scratch.' This strategic pivot suggests a strong emphasis on optimizing operational costs and leveraging the most performant AI tools available to enhance Replit's cloud-based software development platform.
Amjad Masad
Amjad Masad's public statement at TechCrunch Disrupt introduced a layer of complexity to Replit's AI strategy. While he clearly articulated the company's rationale for moving away from Anthropic, citing cost, partnership dynamics, and performance, this declaration was met with skepticism from sources close to the situation. These insiders, who requested anonymity, asserted that the partnership between Replit and Anthropic remained active. This discrepancy creates an intriguing narrative, questioning the timing and completeness of Masad's disclosure and highlighting the often opaque nature of corporate AI collaborations.
Claude Fable 5
The claims made by Replit's CEO were directly challenged by recent evidence of the company's continued engagement with Anthropic's models. As recently as a few weeks prior to Masad's announcement, Replit's own website, blog posts, and social media activity indicated active use of Anthropic's products. Specifically, a late September blog post cited Anthropic's Claude Fable 5 and Claude Fable 5.1 as models actively running in Replit Agent's production environment. Furthermore, a Replit product employee promoted the capabilities of Anthropic's flagship model, Opus 5.5, on X around the same time. These instances suggest that the transition away from Anthropic, if indeed underway, was either very recent or not fully implemented at the time of the CEO's statement, leading to confusion.
15%
Beyond the specifics of the Anthropic partnership, Masad also provided insights into Replit's broader business trajectory and its commitment to AI safety. He noted a significant shift in the company's revenue streams, with enterprise contracts growing from 2% of revenue last year to 15% this year, and B2B revenue now exceeding 50%. This indicates a successful expansion into the enterprise market, diversifying Replit's user base beyond individual developers and startup founders. Additionally, Masad emphasized the critical importance of robust AI safety measures, likening the need for a secure 'habitat' for AI agents to the optimal working environment required by humans. He detailed components of this habitat, including good orchestration, error recovery, security, sandboxing, networking, identity, and authorization, underscoring Replit's focus on responsible AI deployment.
Key points
- Replit CEO Amjad Masad announced the company is no longer working with Anthropic.
- Reasons cited include cost, partnership, and better performance from OpenAI's models.
- Insiders familiar with the partnership dispute Masad's claim, stating the collaboration is ongoing.
- Recent evidence, including blog posts and social media, shows Replit using Anthropic models weeks ago.
- Replit's enterprise revenue has grown significantly, now accounting for 15% of total revenue.
Replit's shift to OpenAI could lead to more efficient and powerful AI coding tools, potentially accelerating development for its users and strengthening OpenAI's position in the developer ecosystem.
The conflicting reports about the Anthropic partnership could create confusion and distrust regarding Replit's strategic decisions and transparency, potentially impacting its reputation or future collaborations.
