Just remember to benchmark it yourself first with you private task collection, so you can actually measure them against each other. Pretty much any public benchmark is unreliable at this moment, and making model choices based on other's benchmarks is bound to leave you disappointed.
This. Last benchmarks of DSv3.2spe hinted at beating basically everything, yet in my testing even sonnet is miles ahead both in terms of speed and accuracy
I still need to talk very fast to actually chat with ChatGPT which is annoying. You can tell they didn't fix this based on how fast they are talking in the demo.
"If opportunities for role fulfillment fall far short of the demand by those capable of fulfilling roles, and having expectancies to do so, only violence and disruption of social organization can follow."
