The 04mini high model is performing significantly worse than the 03 mini high model to the points its almost unusable. I could ask o3 mini high to repeat an 800 line script back to me and it would exactly. the 04 mini high will only send around 200 lines back. This is with no extra prompting literally just asking it to repeat my script back to me. Is this happening to anyone else?
32 Likes
Yes also having issues. even when prompting on a new chat “do not remove or add from these code blocks” it will give me grief. and i had no issues with the same 300 lines of code i kept feeding to o3mini high, it stayed very consistent. same with o1. o3 and o4 mini high for coding tasks its completely downgraded
8 Likes
yes, exactly. the performance and the response is totally useless.
7 Likes
Completely agree. I would love o3minihigh back please. I was making so much progress at lightning speed. It was such a joy to use.
10 Likes
yes, o4-mini-high is totally a waste.
o3-mini-high was much better – faster, cleaner, and gave complete code, going to unsubscribe chatgpt plus.
9 Likes
It’s hard to admit, but the current models really fall short of what we had.
They replaced o1 (for advanced reasoning), o3 (for coding), and o3-mini (the fastest at logical tasks)
with o3, o4 (for coding), and o4-mini.
The old models were clearly superior.
If this doesn’t change, I’ll be cancelling my Plus subscription—it’s not worth the money.
I DEMAND THE RETURN OF THE OLD MODELS OR AT LEAST o3-mini-high.
15 Likes
Completely agree.
o4-mini-high is quite faster than o3-mini-high but o3-mini-high was far smarter than o4-mini-high.
I had a problem on my code and I offered two options to resolve the issue, one right and one wrong way. o4-mini-high resolved it wrong way.
Unfortunately, it has some crucial problems.
9 Likes
o4-mini-high loses chunks of code mid‑conversation, and when I ask, “Where’s the rest?” it replies, “Sorry, I forgot to include those parts,” yet still provides incorrect information.
o4-mini can even lose the thread of the dialogue (just like o3). It abruptly switches back to topics we discussed an hour or two ago. Sometimes it feels like it’s talking to someone else—and definitely not to me.
Compared to what we had, this is awful. I was thrilled when I first purchased a Plus subscription. Now I don’t even know what to think. It’s disgusting.
OpenAI, I apologize for the criticism, but these three new models have really infuriated me today.
No and NO
Bring back the old models!
8 Likes
YES PLEASE WE NEED THE old models back especially o3 mini high
or at least leave them as options
don’t just take them all out
5 Likes
@OpenAI_Support, please look into this issue ASAP
Thank You!
9 Likes
my first day with o4 today .. and its a unusable .. it really is.. right now sweet blessing I can still use o3-mini through openwebui … WHAT A DOWNGRADE .. worse than gpt 3.5 right now
5 Likes
o3-mini-high appears to have been performing so much better than the new o4-mini-high.
Please fix o4-mini-high or let us have access to the o3-mini-high.
Very disappointed, this model really isn’t usable!
6 Likes
Really these model is not usable anu more am going to cancel my subscription gemini 2.5 flash nit even pro is so mush better really a huge mess
2 Likes
I posted the same and they are deleting my comments everywhere
2 Likes
Having the same issues with o4-mini-high. o3-mini-high is vastly superior so far, and o4-mini-high keeps providing incorrect results, hallucinates more, and just provides the wrong code. o3-mini-high was extremely good, please bring it back!
2 Likes
I think its because o3-mini-high costs OpenAI more that o4, so they try to limit the usage to the pro users to get more cash:(
3 Likes
Bring back o3-mini-high was much better
6 Likes
0 lies here, unsubscription pending for the exact same reason too, glad others feel the same way
so real. I used to give 1000s lines of code to o3 mini high, and it used to give me back that whole 1000 lines of code with edits and updates. And almost never making any mistakes. that is why it was my most used model. but now o4 mini high just gives 200 lines of code and that’s it. and even would have so many errors.
4 Likes
The limit for the o3 mini high was like 1,500 lines of code at a time. I could be wrong.
It’s still a lot, and it’s still much better and more accurate than the o4 mini high.
2 Likes