News for hackers. Stuff that matters.

A comment on Gemini 2.5. Parent comment

Re:Gemini 2.5 by bangaladore on Tuesday March 25, 2025 @01:43PM (#43473923)
Reasoning was supposed to be that for "Open" AI, that's why they go to such lengths to hide the reasoning output. Look how that turned out.

Right now, in my opinion, OpenAI has actually a useful deep research feature which I've found nobody else matches. But there is no moat to be seen there.

[ Parent | on HN ]
Re:Gemini 2.5 by CamperBob2 on Tuesday March 25, 2025 @01:49PM (#43473981)
If you've seen DeepSeek R1's <think> output, you'll understand why OpenAI hides their own. It can be pretty "unsafe" relative to their squeaky-clean public image.
[ Parent | on HN ]
Re:Gemini 2.5 by stepanhruda on Tuesday March 25, 2025 @07:13PM (#43477094)
They don’t hide reasoning output anymore?
[ Parent | on HN ]
Re:Gemini 2.5 by bangaladore on Thursday March 27, 2025 @01:12PM (#43495677)
I was looking at this the other day. I'm pretty sure OpenAI run the internal reasoning into a model that purges the reasoning and makes it worse to train other models from.

I might be mistaken, but originally the reasoning was fully hidden? Or maybe it was just far more aggressively purged. I agree that today the reasoning output seems higher quality then originally.

[ Parent | on HN ]