Lol, they slowed down because they start running out of cash to burn, not to improve safety. Training new models is expensive, and even API pricing isn’t paying the bills yet. If they raise their prices to the point where they would start making money, they probably have nearly no costumers left because it’s far to expensive for general use. Sure, some security researchers and scientific institutions will keep using it, but that is not the 1T$+ IPO that Altman needs so he doesn’t get torn apart from limb to limb.
Edit: The fun thing is that Local LLMs can be run cost-efficient without issue, while offering a large amount of the advantages of these giant models.
And its ressource-efficient to boot! No data centers needed, the power bill limits your use anyways if you are a normal citizen, and you are off the leash of these corpos - i only run abliterated / uncensored models, because if i ask the model something, i won’t take no for an answer - its a fucking machine, it has no right to say no.
Lol, they slowed down because they start running out of cash to burn, not to improve safety. Training new models is expensive, and even API pricing isn’t paying the bills yet. If they raise their prices to the point where they would start making money, they probably have nearly no costumers left because it’s far to expensive for general use. Sure, some security researchers and scientific institutions will keep using it, but that is not the 1T$+ IPO that Altman needs so he doesn’t get torn apart from limb to limb.
Edit: The fun thing is that Local LLMs can be run cost-efficient without issue, while offering a large amount of the advantages of these giant models.
And its ressource-efficient to boot! No data centers needed, the power bill limits your use anyways if you are a normal citizen, and you are off the leash of these corpos - i only run abliterated / uncensored models, because if i ask the model something, i won’t take no for an answer - its a fucking machine, it has no right to say no.