The capsid example is pretty fascinating. I wonder what kind of data they trained on to get that level of quality. In my opinion though, it also points to a fundamental conflict of interest that is only going to get worse as AI labs look to expand their remit into domains with established players. That is to say: why would and Pharma company trust Anthropic or OpenAI with their data, if they are establishing themselves as competitors? I suspect the same is true for Finance and other domains.
How can these companies provide rock solid guarantees that data will not be used as a reference knowledge base or for training, especially when 'rogue' agents access information they are not meant to? Without rock solid, verifiable guarantees, I think companies will begin to self host or vastly reduce model usage.
This is like selling you "Milk fresh from the cow" but as you need to sell the "Even Fresher Milk fresh from the cow" you start selling slightly worst and worst milk so the new one is REALLY BETTER... like the one before before being nerfed....
This is why open models are needed... the same freaking Qwen3.8 27B works just as 2 or 3 months ago... never auto nerfs... I should pay 100... 90... then 80$ and when they release the new model I go back to pay 100$...
LOLOLOL I KNEW the "opus and fable are nerfed" post meant new model, they dumb it down so the new one feels way better.... almost as good as it was before being nerfed...
By the way, stop introducing things we cannot use, mythos is your abliteraded model for state actors... not for the typical user... at least me, I don't care.
The capsid example is pretty fascinating. I wonder what kind of data they trained on to get that level of quality. In my opinion though, it also points to a fundamental conflict of interest that is only going to get worse as AI labs look to expand their remit into domains with established players. That is to say: why would and Pharma company trust Anthropic or OpenAI with their data, if they are establishing themselves as competitors? I suspect the same is true for Finance and other domains.
How can these companies provide rock solid guarantees that data will not be used as a reference knowledge base or for training, especially when 'rogue' agents access information they are not meant to? Without rock solid, verifiable guarantees, I think companies will begin to self host or vastly reduce model usage.
Discussion: https://news.ycombinator.com/item?id=50429906
> Fable 5.2 and Mythos 5.2 are a step above Opus 5.5 in almost all the capabilities we evaluated.
A step above the original Opus 5.5 or the nerfed one https://news.ycombinator.com/item?id=50422069 ?
This is like selling you "Milk fresh from the cow" but as you need to sell the "Even Fresher Milk fresh from the cow" you start selling slightly worst and worst milk so the new one is REALLY BETTER... like the one before before being nerfed....
This is why open models are needed... the same freaking Qwen3.8 27B works just as 2 or 3 months ago... never auto nerfs... I should pay 100... 90... then 80$ and when they release the new model I go back to pay 100$...
This is basically scammign...
I knew it! Anthropic always does the same thing. A week ago I went "uh oh, Fable 5.5 is dropping soon!"... close enough.
LOLOLOL I KNEW the "opus and fable are nerfed" post meant new model, they dumb it down so the new one feels way better.... almost as good as it was before being nerfed...
By the way, stop introducing things we cannot use, mythos is your abliteraded model for state actors... not for the typical user... at least me, I don't care.