Wanted to make a quick demo for the AuraSR-V2 model from fal.ai using fastAPI and HTML/HTMX, but after using Sonnet 3.5, GPT-4(o), and Deepseek Coder none of them were able to get anything working the way that I wanted it. The main issue was sending and receiving the image to and from the backend. I even went and found some JS libraries for it to use to help it, but it did not work. In this case I think I could have made it faster myself, or if I had only used the LLMs for parts of the code instead of trying to have it make the whole thing. I also think it would have done better if I had used React, but for a one page project like this React is overkill. Check out the code here (or, more importantly, the prompt). Hopefully I can get this working tomorrow.