Comparative test of 9 AI engines with the same prompt: build a modern website...
I tested the exact same prompt across nine GenAI engines. Here’s the lineup:
The verdict
Lovable is loveable. Bolt had a lot of issues that needed tweaking. App builders like Lovable and Bolt create React applications that need to be built in order to get the HTML/CSS and JS files. These websites are actually more like apps, fully relying on JavaScript. It's probably an overkill for an average website, but very powerful when creating interactive web applications. Just to make things clear: Lovable and Bolt use GPT-5 in the back! So we're kind of comparing apples with oranges.
ChatGPT-5 is the best coding assistant today. It nailed every requirement on the very first try—no glitches, no extra steps. Even better, it delivered a complete zip file with all files and folders ready to go.
Gemini, Grok and Claude performed reasonably well but forced me to manually create the project structure.
Perplexity demanded lots of fine-tuning and tweaking. It used a jQuery lightbox-plugin without referencing jQuery in the code.
Deepseek had a very hard time. DeepSeek constantly throw me a "Server busy, please try again later"-error. But it finally came through. DeepSeek's attempt shows a lot of issues though.
And Copilot? It basically stalled - making it useless for coding tasks of this kind.
Conclusion
As an experienced developer, I don't rely on AI app or website builders like Lovable and Bolt, since I can handle hosting, testing, and security myself. However, for newcomers, these tools can be quite helpful—though you'll still need to know how to trace and report bugs, because let's face it, there are plenty of them.
That said, even seasoned developers can't match the speed at which advanced GenAI tools like GPT-5 and Gemini can churn out websites and web applications. The more refined your prompt, the more impressive the output. This holds true for all GenAI coding assistants.
One thing's clear: a new era is unfolding. In the near future, AI-generated websites, web apps, and standalone applications will become the norm, regardless of whether they are built by experienced developers. As a user, you won’t be able to tell the difference between something generated by AI or handcoded.
Right now however, in 2025, GenAI code generators just aren’t in the same league as big developer ecosystems like WordPress, Drupal, or Joomla. Those communities of real people are still way ahead, building frameworks that are far more solid than anything AI can spit out today. But who knows what the next five or ten years will bring? Maybe by then, GenAI will be able to whip up software that completely outshines WordPress and its thousands of themes and plugins. Or maybe not. It all depends on where GenAI companies decide to put their energy. Will they chase image, sound, and video generation? Or will they double down on creating powerful business applications that change the game? My bet is that GenAI will go wherever the money leads.
