⏳ HOURGLASS JUST LEVELED UP ⏳
We shipped the biggest upgrade to https://t.co/qZVsU4f9TX since launch. Everything below is live on the site RIGHT NOW. 👇
@dhrumiliso31702 @kimmonismus it's literally just better training. GPT4 was 1.7T parameters and Opus 4.6 is roughly the same. the difference in capability, intelligence, understanding between the two is night and day. It's an order of magnitude at every task. thats what 2 years brings
@roeytechai@kimmonismus yes? Local A3B models are drastically more effective and efficient running off 16GB VRAM and a bean, than even GPT 3.5 and 4.0 were just 1 year ago... running off a heap of H100's