Key facts
- Base44, a vibe coding platform acquired by Wix for $80 million, has launched its own AI model named Base1.
- The custom LLM aims to improve latency, cost, and efficiency compared to general frontier models.
- Base44's strategy focuses on defensibility by owning its model, data, and infrastructure.
- The model was trained on data from millions of real user interactions on the platform.
- Cost reduction for customers and improved margins for Base44 are expected benefits.
Base44, a vibe coding platform acquired by Wix for $80 million just one year ago, has begun rolling out its own AI model, Base1. This strategic move addresses the growing debate within the AI community about the long-term defensibility of startups built on third-party foundational models and whether general-purpose frontier models are optimal for all use cases.
According to Base44 founder Maor Shlomo, training and owning the model allows for significant optimizations in latency, cost, and efficiency. While competitors like Swedish startup Lovable rely on external LLMs, Shlomo anticipates that other scaled players will also develop their own models. Jonathan Userovici, a general partner at VC firm Headline, identifies data, distribution, and tech stack as key ingredients for AI startup defensibility, a pattern Base44 appears to be following.
The Base1 model was developed and trained on a dataset derived from "tens of millions of real user interactions on the platform." This approach aims to provide a competitive edge against both other specialized startups and larger AI labs like xAI and Anthropic, which are also entering the app creation space. Shlomo believes that specialized models will outperform general ones for specific tasks.
Userovici, however, cautions that frontier models should not be underestimated, pointing to Harvey's decision to abandon its own model training. He frames Base44's move within a broader trend driven by rising inference costs, which enterprise customers are increasingly concerned about. The demand is for orchestration and optimization solutions that maintain performance while controlling costs.
While enterprise clients are a minority for vibe coding platforms, they represent a growing revenue share. Base44's decision to develop its own LLM is likely motivated by cost reduction, better alignment with user preferences, and improved speed. The company expects that owning the model will give it direct control over compute and inference spending, leading to a stronger margin profile over time. This move comes as Base44 has been growing its headcount since the acquisition and recently surpassed $100 million in annual recurring revenue, though still trailing competitor Lovable's $500 million ARR.