Tried Google's new Gemini model against ChatGPT for coding last week

I was building a small web tool to organize my recipes and decided to test both AI assistants side by side. Google's Gemini 2.0 handled the whole thing in one shot without any errors, while ChatGPT 4o kept giving me broken code that needed fixing. The difference was Gemini caught my vague instructions better and filled in the gaps on its own. Has anyone else noticed this gap with more complex tasks like building actual working apps?

2 comments

2 Comments

johnthompson1mo agoProlific Poster

You know a buddy of mine tried this exact thing building a little dashboard for his fantasy football stats. He's not a coder at all, just messing around on a weekend. Said ChatGPT gave him a Vue setup with a bunch of missing dependencies and he couldn't get it to start. Gemini just gave him one html file with everything inline, worked right away when he opened it in a browser. Kinda weird how different they handle the same ask.

oliver_mitchell1mo ago

I built a recipe app too last month and Gemini totally crushed it compared to ChatGPT for me. My instructions were pretty vague like "make a search bar that filters by ingredient" and Gemini just figured out the whole logic without me spelling it out. ChatGPT kept giving me these broken imports and missing pieces that I had to debug for an hour. Maybe it's just me but Gemini seems way better at reading between the lines on these bigger projects.