Google’s Gemini 3 is here — and it’s already beating ChatGPT in key benchmarks
6mon 25m ago by lemmy.world/u/jaykrown in AINews from www.tomsguide.com
Is it easier to not use? That's my most important benchmark when it comes to llms
That's what the benchmarks attempt to find out, I think it depends on what you're trying to do as well. Easier to use for what?
Which one is the easiest not to use. Llm are catastrophic for the environment
If all goes well: nothing
It seems to have regressed vs Gemini 2.5 in some long context comprehension, like asking stuff about papers or stories... Which is basically the only thing I use Gemini for, since open/local models are so good at shorter contexts now.
This isn't suprising. For that stuff, Gemini's peak was somewhere in the 2.0/2.5 previews, but then they deep-fried it to benchmaxx coding and lm-arena.