What happened
Replit CEO Amjad Masad announced on stage at TechCrunch Disrupt that Replit is no longer working with Anthropic, citing cost, partnership dynamics, and performance metrics favoring OpenAI models. However, the report notes that insiders familiar with the partnership claim it is still in place, and a September blog post from Replit cites Anthropic's Claude models as running in its production environment. TechCrunch has reached out to Replit for clarification.
Replit CEO and co-founder Amjad Masad stated during a keynote at TechCrunch Disrupt that his company is no longer working with Anthropic. Masad attributed the decision to a combination of factors, including cost, the nature of the partnership, and performance metrics. He specifically noted that Replit uses internal benchmarks to assess performance and that OpenAI's models, specifically Sol and Astra, perform better on building vibe-coded applications from scratch according to their public benchmarks.
The announcement contradicts previous indications of a continuing relationship. Replit had announced a partnership with OpenAI in August, but there had been no prior public indication that the Anthropic partnership had ended. Under the previous arrangement, Replit used Anthropic's Claude models to power its cloud-based software development platform, a relationship that was credited with helping the company grow its revenue significantly.
Despite the CEO's statement, the TechCrunch report notes that the claim is disputed. One insider familiar with the partnership, who requested anonymity, stated that the partnership was still in place. Furthermore, the article points to concrete evidence suggesting continued use of Anthropic products as recently as September, citing a Replit blog post that listed Anthropic's Claude Fable 5 and Claude Fable 5.1 as models running in Replit Agent's production environment.
TechCrunch has reached out to Replit for clarification and stated it will update the story if a response is received. The report also includes other comments from Masad regarding Replit's business growth, noting that enterprise contracts have grown from 2% of revenue last year to 15% this year, with over 50% of total revenue now coming from B2B sources. Masad also discussed the importance of creating a safe 'habitat' for AI agents, emphasizing the need for good orchestration, error recovery, and sandboxing.
Source details: techcrunch.com ↗
Why it matters
This potential shift in AI infrastructure partnerships highlights the competitive volatility in the enterprise AI market. If Replit has indeed switched primary reliance from Anthropic to OpenAI, it signals a significant realignment in how major coding platforms source their underlying models. The discrepancy between the CEO's public statement and insider reports underscores the opacity of these commercial relationships and the rapid pace at which AI vendors are being evaluated and swapped based on cost and performance benchmarks.
The potential end of the Replit-Anthropic partnership is significant because Replit is a major player in the AI coding space, and its choice of underlying models influences the capabilities and costs for its large user base of developers and startups. A shift to OpenAI models could indicate a broader trend of enterprises prioritizing specific performance benchmarks or cost structures over long-term vendor loyalty.
The contradiction between the CEO's public statement and insider reports highlights the complexity of AI vendor relationships. It suggests that these partnerships may be more fluid or multi-layered than publicly perceived, with companies potentially using multiple AI providers simultaneously or transitioning gradually rather than making a hard cut. This opacity makes it difficult for customers and investors to fully understand the technical foundations of AI products.
This event underscores the intense competition among AI labs for enterprise contracts. The fact that Replit cites performance benchmarks as a key factor in its decision-making process reflects the maturing of the AI market, where technical merit and cost-efficiency are becoming primary drivers for adoption, rather than just early access or brand recognition.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
crm_get_transaction(id='4092').Which component of an AI application is the machine-learning model itself?
What to watch next
Watch for an official statement from Replit or Anthropic clarifying the status of their partnership. Additionally, monitor Replit's product updates to see if there are visible changes in model availability or performance that align with a switch to OpenAI models, and observe if other major AI coding platforms announce similar shifts in their vendor relationships.
The most immediate next step is an official clarification from Replit or Anthropic. If Replit confirms the switch, it may provide details on the timeline and the specific models now powering its platform. If they deny it, it will raise questions about the accuracy of the CEO's remarks and the nature of the ongoing relationship.
Monitor Replit's developer documentation and release notes for changes in model availability. If OpenAI models become the default or primary option for new features, it would corroborate the CEO's statement. Conversely, if Claude models remain prominent, it would support the insider's claim that the partnership is still active.
Observe how other AI coding platforms and enterprise software companies respond to this news. If other major players also announce shifts in their AI vendor partnerships, it could signal a broader industry trend of diversifying or switching AI providers based on performance and cost metrics.