Base44 introduced its own AI model to compete with tools from Lovable, Replit, and Cursor.

Founder and CEO Maor Shlomo explained to Business Insider that the San Francisco-based vibe-coding company developed its LLM specifically to move beyond the repetitive AI-slop aesthetic common in many generated products.

Experts note that sites created with top models such as Anthropic's Opus 4.8, OpenAI's GPT-5.5, or Gemini 3 frequently feature matching layouts and elements including rounded corners, neutral beige tones, and heavy emoji use. Rivals like Lovable and Replit rely on these same base models.

Paul Bakaus, CEO of AI design firm Impeccable, described such outputs in a June interview as resembling an algorithmic Uniqlo or Ikea: practical yet plain.

Shlomo noted that future iterations of Base 1 will generate distinctly original appearances each time, though that capability still requires further development.

He added that the model would deliver results more quickly and consume fewer credits than frontier alternatives. Base44 operates as a subsidiary of website builder Wix.

To evaluate performance, the same website prompt was run on both Base 1 and Anthropic's Opus 4.8.

Making a website for a fictional mood lamp

The task involved creating an e-commerce page for a made-up emotional support lamp that syncs with Google Calendar and shifts colors based on events. Yellow signaled focus time, blue indicated downtime, pink marked meetings, and red warned of approaching deadlines.

The prompt was submitted to Base44 without any design instructions to observe the model's independent choices. Base 1 was selected as the model, an option available only to paid Builder subscribers at forty dollars monthly.

After several minutes, Base44 generated the Lumos site.

Base 1 vs Opus 4.8

The initial design stood out with a deep blue background and bright yellow accents, differing from the lighter palettes typical of many AI-generated sites. Rounded corners and frequent emojis remained visible as standard vibe-coded traits.

Shlomo acknowledged that Base 1 still needs refinement to achieve truly distinctive results.

The identical prompt was then run using Anthropic's Opus 4.8. Generation took noticeably longer, roughly twice the time required by Base 1. The output featured the expected beige background.

Other elements closely matched the Base 1 version. Base 1 completed the task faster while both consumed 1.2 message credits from the monthly allowance of 250.

When asked to add green for exercise and teal for eye breaks, Opus 4.8 used 1.4 credits. Base 1 finished the update with 1.2 credits and in less time.

Although the gap was modest, Opus demonstrated higher credit consumption for equivalent changes.

Overall, the models performed similarly, with the main distinction being that the Opus design appeared slightly more conventional.

Shlomo said the team plans to apply reinforcement learning to encourage Base 1 to produce fresher, more varied designs in upcoming versions.

Further testing will follow the next release.