@karpathy on having models collaborate on complex projects

Another technique for getting more out of LLMs, which each have their own approaches and personalities and quirks. Have more than one perform the same task, then combine the results, and ask them each to highlight the strengths and weaknesses.

I love doing this actually :). I think it's a pretty powerful eval too. Have all models generate something, then put it all together and give it back to all of them and ask them to rank all outputs. I thought models might have a bias to prefer their own outputs, but this doesn't…
— Andrej Karpathy (@karpathy) September 5, 2025

I think congrats again to OpenAI for cooking with GPT-5 Pro. This is the third time I've struggled on something complex/gnarly for an hour on and off with CC, then 5 Pro goes off for 10 minutes and comes back with code that works out of the box. I had CC read the 5 Pro version…
— Andrej Karpathy (@karpathy) September 5, 2025