I stumbled upon an interesting project where someone recreated the Bernoulli demo using their code on Gemini 2.5. What’s impressive is that it only took 30 seconds to generate, compared to the 3-minute demo by ChatGPT. It just goes to show that with the right tech, you can achieve similar results without breaking the bank – or in this case, without a 60-billion-dollar preference stack.
I think it’s fascinating to see individuals pushing the boundaries of what’s possible with AI and language models. It’s a great reminder that innovation can come from anywhere, and you don’t need a massive budget to make something cool happen.
The project’s creator, epreisz, shares their experience and insights on the Reddit thread. If you’re curious about AI, language models, or just want to see what’s possible with Gemini 2.5, it’s definitely worth checking out.