Hacker News new | ask | show | jobs
by make3 58 days ago
if you look at the details of the numbers of the benchmarks that you shared, Sonnet 4.5 crushes gemma 4. Somehow the first link doesn't run Sonnet on the multi modal benchmark, that's why the top score looks close, it beats Gemma at every benchmark they actually ran. The arena in the second shows that it actually destroys Gemma 4 as well, not close
1 comments

The second one is Sonnet 4.6 not 4.5. If you change it to 4.5 Gemma 4 actually beats 4.5