GPT-4o is a bad release requires further tuning

GPT-4o is a bad release, I’ve done extensive testing it’s unstable. Sorry Open AI it needed more tuning, good for novelty use. I always revert to GPT-4 till locked out and then I use GPT-4o. I wonder what it did to the custom GPTs using GPT-4 Architecture.

2 Likes

Not particularly constructive feedback. If you want anyone to take this post seriously, you need to actually describe what you’re trying to do and where it’s failing.

1 Like

True:
What it is, is an expression of frustration after 4-5 hrs of testing on one project . These pitfalls accumulate. I spend more time than the average person on ChatGPT. I use it for Programming, writing for actual court papers files(research), and I tested it side by side during college courses. I’m still an advocate for AI and have tested all AI by big data. GPT still my go to, sometimes Github Copilot. For this reason it’s important I send a signal to the top, to adjust accordingly.

There’s another post where I did extensive ordering of the models from best to worse.

All good.