AI is made by the rich for the rich, mythos remains under lock and key only available to the Pentagon, the other flagship models have all increased the cost of tokens by at least 50% and will only continue to increase the cost of tokens over time as the initial funding phase of AI runs out. The status quo is quickly forming and flagship models are simply deemed out of reach for the average plebian who cant spend between $250,000 - 400,000 a day on tokens for the only model that matters.Either you support the panopticon dystopian hell coming or you don't - Choose a side.
I choose a third option
>>536966940you choose pretending* to be a 3rd option in order to dodge accountability for the things which your fascist capitalists have done to fuck us over, you mean.
Mythos is a scam. Claude has no bullshit underperformed openAI consistently. Codex shits on it for a tenth of the token cost. Their scam right now is "its too dangerous and that is why you can't use it for any biology or chemistry now, uhhh, its not to hide that it isn't any better and cut down costs""uhh are you asking help with math? uhh, maybe you're under 18 lets reduce token cost and block your account temporarily!"Its excuse after excuse... Lets be clear here too, do you honestly believe that anthropic has the research capacity of open AI? Its all ROI. They simply do not have the financial backing. Their model isn't mystical and special. Anthropics goal the last year has been regulatory capture rather than competing with openAI. Given this article is their own reporting, I'm assuming they found some other loophole to give them an advertising out from performing. Something like if they meet X metric (self attestation) they are not to export the model. So now they claim it, there is no USG review of the model, and they report this as if their model is too good, too dangerous, too amazing but you can't see it because it goes to a different school.
>>536966856welcome to the permanent goy underclass
>>536966856>AI is not for (You)I know, it lies about jews.
>>536966856Well I mean I have a couple agents, they’re enjoyable. I’m not gonna worry about all that.
damngood thing my computer has infinite tokens,huh didn't think about that one dumbass?>b-b-but their super pooper moldes are super dipper bettererrerr!!yeah and a year later a local model that's 99% as capable with a 120Gb size drops for free
>>536966940Free and open-source models for everyone!
>>536966856GemmaOP BTFO
>>536967276you're right but also codex on xhigh is a bit bad. Pro mogs so hard that it feels like a whole different model. I've almost quit using codex because it keeps breaking my code or making it suboptimal (brittle JAX code that needs to stay at 99% device util at all times) and have defaulted to manually prompting pro like the good ol times. The things that you would get in four hours via the codex cli tend to happen in an hour via pro and you can leech like a $5k of token-equivalent by using up the weekly pro prompts that you get from the $200 sub. Codex is mostly good for running tests while you're asleep.
>>536966856>our thing is so awesome, the government banned it>but you can still use it for premium price ;)desperation is a stinky cologne anthropic
>polymorphic orthogonality
>>536966856im using it right now. Nothing was taken down. lol
>>536966856in 10 years you will have easy access to models that will make fable and everything else look beyond infantile
