I just noticed that the O1-Pro version, which was previously the best AI for coding at $200 per month, has drastically declined in quality. It used to give thorough responses, especially on complex queries, but now it's producing short and less informative replies. While it still seems competent at debugging, I can't understand why there's been such a sudden drop in capability. How can a premium service just decline without any notice or explanation?
5 Answers
I've definitely noticed the lack of communication from them. O1-Pro was the go-to for fixing bugs that others couldn't touch. I tested it recently, and while it’s still decent, the quality just isn’t what it used to be. Hope they're not permanently downgrading it!
Yeah, I’m still hoping this leads to an O3 release. That would be nice!
I switched to O3 right after I noticed O1-Pro's responses dropped. O1 was stellar for a while, but post-O3 release, it felt ‘nerfed.’ Switching was the best decision for me.
Even with O3, I find O1 sometimes had better reasoning for complex problems.
It costs them a lot to keep O1-Pro running. The $200 subscription definitely doesn’t cover the usage. If you want the best responses, you might need to switch to the API, but be prepared to spend more.
Is there actual proof that it costs them this much? Seems off.
It's disheartening to see how little they've communicated with us. It feels like they're just squeezing users now.
Honestly, it feels like a classic strategy to grab market share and then cut quality to boost profits. They're just interested in keeping investors happy now.
Couldn't agree more. It's all about the cash flow for them.
I'm checking out Deepseek now because of this. Can't trust them anymore.
It turns out O1-Pro was already marked for deprecation due to O3's release. This seems like the end phase to save resources for upcoming launches. It's frustrating as a subscriber, since they could have communicated this more clearly!
Totally get that. It could have been an easy announcement instead of leaving us in the dark.
Yeah, if they called it a downgrade at least we’d know what to expect.
I tried it on a complex project recently, and it did manage to spit out a decent 1200-word response, so maybe it’s not entirely gone yet.