You can host a small LLM model with a phone, or a whisper or a TTS, maybe some light diffusion.
You cannot do tool use, reasonable coding, get non-hallucination answers on even moderate questions. That stuff starts to get viable around 24G and really needs closer to 200G to be able to do most of that they’re doing.
the reason they bought up all the ram is to keep someone from buying 200G of vram and diying cards putting them out of business.
I read this as about general social media, not LLMs.
LLMs, I agree the at-home hardware is pricey and not super competitive right now. Plenty of frontier models that are kept from the public, with VC-discounted rates because the IPO hasn’t dropped yet. (If you’re expecting a market crash, this is going to get much better fairly soon. In the mean-time, subscriptions are probably more cost effective than just buying the hardware at old prices would have been.)
Frankly, you can self-host most of these services on an old phone. Setting it up is getting easier than ever with cheap agents flying around.
You can host a small LLM model with a phone, or a whisper or a TTS, maybe some light diffusion.
You cannot do tool use, reasonable coding, get non-hallucination answers on even moderate questions. That stuff starts to get viable around 24G and really needs closer to 200G to be able to do most of that they’re doing.
the reason they bought up all the ram is to keep someone from buying 200G of vram and diying cards putting them out of business.
I read this as about general social media, not LLMs.
LLMs, I agree the at-home hardware is pricey and not super competitive right now. Plenty of frontier models that are kept from the public, with VC-discounted rates because the IPO hasn’t dropped yet. (If you’re expecting a market crash, this is going to get much better fairly soon. In the mean-time, subscriptions are probably more cost effective than just buying the hardware at old prices would have been.)