Gemini 3.5 Flash beats Opus 4.8 on bluffbench(bsky.app)

2 pointsby ionychalletzten Monat2 comments

Gemini 3.5 Flash beats Opus 4.8 on bluffbench

https://bsky.app/profile/simonpcouch.com/post/3mmwroep6lc2y

2 comments

Too bad Google’s internal culture and politics keep ruining their products. You can’t keep rebranding things and deprecating APIs and all that. People won’t rely on it. And they aren’t. Google keeps screwing up something they should have won at. Maybe their anti competitive power in distribution will save them, but there’s basically no consumer or business interest in Gemini. And I don’t think benchmark model performance will change that.

The real issue is that Google has to purge most of its senior leadership and bring in better, hungrier talent that can lead the company through all this. Right now it looks like it’ll become the next IBM.

youarenotyuletzten Monat

idk why but gemini acts very well on benchmarks, but it's clearly falling behind gpt and opus in actual tasks