The Base44 Base 1 model, launched on 29 June, is the San Francisco startup’s attempt to solve one of vibe coding’s most visible problems: the generic, interchangeable look that frontier AI models keep producing when they build websites and apps.
Base44 CEO Shlomo announced the model directly to Business Insider, explaining that users can now select Base 1 from a menu of available models on the platform, alongside Claude’s Opus 4.8, Fable 5, and OpenAI’s GPT-5.5. The company had been working on it for roughly six months but reached breakthroughs in recent weeks that allowed an earlier-than-planned release.
How the Base44 Base 1 Model Was Built
The design problem Shlomo is trying to fix is straightforward. ‘Everybody feels like they’re getting the same UI when they’re coding with the general models,’ he told Business Insider. Frontier models, he argued, must be competent across everything from poetry to code. ‘And we think that if we take a model and we squeeze its ability to be really, really good at one use case, then we have a shot,’ he said.
The key ingredient is data. Base44 sits inside SiliconANGLE-covered acquirer Wix, which brings a large in-house design team and the design data that generates. AI Chat Daily reports that Base 1 was trained on a dataset built from tens of millions of real user interactions on the platform. The team is running reinforcement learning on the model, prompting it to keep generating designs that look new and distinct each time.
Shlomo acknowledged Base 1 is ‘not yet there,’ but his stated aim is for it to ‘create something that looks uniquely different’ with every generated interface. Paul Bakaus, CEO of AI design startup Impeccable, described the standard vibe-coded aesthetic in a June interview with Andreessen Horowitz as an ‘algorithmic Uniqlo or Ikea’: not bad, but not unique. Telltale signs, he said, include beige or tinted backgrounds and sans-serif fonts.
The Base44 Base 1 model launch also carries a margin dimension. Startup Fortune describes it as ‘both a margin play and a competitive signal’, reporting that Base44 was clocking $150 million in annual recurring revenue as of May 2026. AI Chat Daily puts the ARR milestone at $100 million crossed a few months before the launch; where the two figures diverge, Startup Fortune’s May 2026 date makes it the more recent of the two reported figures.
A Crowded Market, a Harder Competitive Threat
Base44 sits among a cluster of vibe-coding rivals including Lovable, Replit, and Cursor. Lovable reached unicorn status in its Series A round last summer and, like most of the field, relies on external large language models rather than a proprietary one, according to TechCrunch. Owning the model stack, Shlomo argues, is what separates Base44 from that group.
TechCrunch also notes that the deeper competitive risk may not come from those peers at all, but from frontier AI labs moving into the space. Cursor and xAI, the parent of Grok, now both belong to SpaceX, and Claude Code has become a vibe-coding player in its own right.
The Wix acquisition that gave Base44 its data advantage closed at $80 million, with additional earn-out payments running through 2029, per SiliconANGLE. At that point, notable customers included eToro Group and SimilarWeb. According to Enterprise Management Associates, Base44 had reached 250,000 users and was already profitable when Wix bought it. TechCrunch reported the startup was approximately six months old and had a team of eight employees at that point.
Whether Base 1 can reliably deliver visually distinct outputs at scale remains the test. Reinforcement learning on a design-specific dataset is the mechanism Base44 is betting on. The next marker to watch is how the model’s output quality evolves as the team iterates beyond those early breakthroughs.