[basch]

Did you look at the charts in the article?

It out-performed every model that wasnt a max/ultrafrontier of some sort, except for the one that the article was extolling the virtues of, including grok high. you could make a good argument that deepseek is a better value, but gemini flash is when bundled is already pretty accessible.

nowhere did i claim that flash was better than fable or 5.5xhigh.

[joe_mamba]

>Did you look at the charts in the article?

I don't care about someone else's charts, i care about my own lived experiences. Benchmarks can be gamed to get to the top of charts. When I pay for a service I care about how it performs in my test cases, not about which tops some random charts.

Read my comment again please. I think I was pretty clear with detailed examples on where Gemini sucks and where it's good at.

>nowhere did i claim that flash was better than fable or 5.5xhigh.

And nowhere did I claim that. I said even basic GPT and Grok are better than Gemini Flash at reasoning tasks. Again, read my comment again, I have already explained why with examples.

[torben-friis]

If you don't care about someone else's charts, why do you expect others to care about your comment?

[joe_mamba]

>If you don't care about someone else's charts

Bruh, do you know what a personal opinion is? You don't need to care about it, and I never said I don't care about other people's opinions, I just said that chart which is based on non peer reviewed information, doesn't match my experience so without further peer reviewed proof to back it up, I don't care about it as my expire shows otherwise.

Would you believe any graph that tells you the sky is brown when you go outside and see that it isn't?