#Does anyone use GPT 5.2 or 5.3-codex as

1 messages · Page 1 of 1 (latest)

coarse dagger
#

I started using it as a main model a few days ago.
Quite good. But I never tried Opus so can't compare.
Befoer I was running Gemini 3 Pro

waxen lark
#

I've tried both and I run 5.3 codex as my main model. I found that 5.3 codex is a lot faster but a lot less fun conversationally. For me, I don't mind that trade off and would prefer speed. I didn't notice a huge difference in tool calls between the two models but I would expect 5.3 to be better.

quick surge
#

I ran codex for a week and had no complains... laugh when the Lev interview mentioned it was german 😄 other than the small hit in personallity works really well.

tame bloom
#

As a German, I can actually confirm that. When Peter said in the interview with Lex that Codex communicates more like a German — concise, fairly cool, not overly enthusiastic — that resonated with me. And that’s perfectly fine for me, because it handles the tasks it’s given very reliably.

I think it’s not particularly strong for very simple tasks, but when it comes to complex problems, it’s clearly much better. I especially like using Codex for programming or refactoring code.

ornate hinge
#

I'm finding the weekly limit has gone down way to fast to make it useful I have 5.2 and down to 57% with some basic chats from yesterday evening