Hang around Hacker News long enough and you’ll inevitably see a thread where people are going back and forth about which foundational LLM is better. Everybody taking their turn hyping their preferred choice while dunking on others.

It’s just boring and tedious at this point.

Your experience with ChatGPT will be different than my experience with ChatGPT based on so many different factors that it is almost impossible to have a meaningful conversation about the rough spots and/or successes we see.

My observation is that all of the models are just asymptotically getting closer to the limit of what generative AI is able to do given the current hardware constraints. Furthermore (at least today), the amount of investment being put into each of them will only guarantee that they will continue leapfrogging each other on some “standard” benchmark every other month.

If you are trying to decide which model to use, I think you should just pick the one that is most convenient to you at the moment (available in your cloud provider, etc.), modularize and isolate the code or business process that is going to interact with the LLM so that you can swap it out later and rest easy knowing you made the right choice.


Originally posted on LinkedIn on January 8, 2026.