have you noticed something new with claude 5 and gpt 6?
this is the first generation where they released the biggest model fable and astra ahead of the smaller ones
this is very interesting and worth speculating a bit
the most reasonable explanation i can think of is that only fable and astra are actually “trained”
opus 5, sonnet 5, and the upcoming gpt 6 sol, terra, luna are all “taught”
this is NOT a good sign. why? have you seen opus 5 and sonnet 5? they are bad
opus 4.6, 4.8 etc were “trained”, because there was never a mythos 4 to teach them, so opus back then was good
it seems a training pipeline where the model is primarily taught by / distilled from a large teacher model have all sort of quirks and are generally not as useful in practice, especially for direct interactive use as a human
perhaps they will nail this down the road, but if they don’t, then what it means to us is -
we will only be able to use fable/astra directly, and need to ask fable/astra to use the other models as subagents. not sure i like it
the first challenge in life is graduating from survival to creative mode
the second challenge is figuring out what to create and how
the third challenge is seeing it to completion without falling back down to survival mode
1/🧠 Sam Altman says ALL jobs are about to be automated by AI.
If you don't want to be POOR, you'll need to train for one of THESE jobs of the future—a thread 🧵
dropping standardized tests while maintaining legacy admissions policies and claiming it's about advancing equality of opportunity is not a serious position.
We’re going to call a flag on this one.
Early this morning a 2011 Black Audi A4 with Quebec licence plate X10 SNP smashed through Vaughan Mills. Two suspects committed a break and enter and fled.
If you have any information on this vehicle or the suspects, please call police.