OpenAI is in a tough spot.

I just canceled my $200/month GPT-Pro subscription. 5.2-Pro is in decline -- it has been getting noticeably worse at a steady rate since introduction. At this point, it's not appreciably better for most queries than Claude 4.5 Opus, and Opus is roughly 10x faster.

▲

the__alchemist 4 hours ago | parent | next [-]

Noticed the same. Also, I noticed the same with the prev version, immediately improved when switching to the new 5.2.

The smoking gun is the time. If I ask it a question that's subtle in "thinking" mode and it starts replying in a few seconds, the answer will probably be trash. I'm almost sure they degrade the models over time.

	▲	IAmGraydon 4 hours ago \| parent [-]
		An interesting approach - start strong when everyone is running benchmarks, taper off through the life of the release, introduce new version that is no better than the last version but it magically seems much better by comparison to the degraded previous version. Rinse, repeat, grift.

▲

weinzierl 4 hours ago | parent | prev | next [-]

My impression is that recently ChatGPT tries to avoid going out to research on the Internet as long as it can. I have to tell it to pull info from the web or verify its answers on the web explicitly.

Could it be that they are trying to save traffic?

	▲	probably_wrong 3 hours ago \| parent [-]
		My non-existent marketing instinct would tell me that they are trying to keep you inside the app to convince you that ChatGPT is the internet, the same way some people wouldn't know there's life outside Facebook. My grumpy instinct tells me they know that they're poisoning the internet and they have given out on trying to weed out the fake websites from the real ones.

▲

franze 4 hours ago | parent | prev | next [-]

Or Opus 4.5 has gotten better since release? Can not point my finger at it but the code I get is most of the time super flawless.

▲

jimbob45 4 hours ago | parent | prev [-]

In favor of what?