Why does deep seek code better than o1 and o3

ChatGPT Use cases and examples
1

I spend hours trying to convert code using open ai models and it’s just a waste of time. Each week it gets worse. Even older open ai features like web searching is now replaced with cache style garbage. Google is back to being better which is 100% open ai fault.

I can type stuff into deep seek and get the answer. Like how open ai was last year with 3.5. No joke it will answer the question and it can give a numerical answer as well which open ai models forgot how to do.

Open ai seems to spend its money on lawyers to lobby and ban better models. If this continues I’m cancelling membership and if open ai keeps getting dumber models I have to cancel.

I pay for this product and it’s becoming more like a charity service.

2

I have looked into this and apparently the open ai model performs so badly now because open ai has loaded it with so many fail safes and ethical rubbish. It even knows this and it knows it’s becoming more useless. Even prompt engineering to make it over work is less effective because it’s been redesigned to detect prompts that might make it more useful. It thinks it’s being hijacked and then goes back to repeating the same thing.

I don’t know how much was spent on making this worse but it’s not worth it. The direction this is going is making a restricted inside the box thinking model that costs a lot and is not the best at programming and maths. All this money could have been spend on making it better and cheaper.