Im sure some of you came across this interesting startup, running 120b parameter models on a tiny 30watt device. This is stage i was waiting for, the only people who like tokens are tho ones selling them well token charge me no more!
Of course nothing will beat a legit data center but this is at the power level and performance level for most of my basic Ai task, cant wait for it to be released
Just wanted to here some options on it
Is there more available than a youtube video?
probably
Their website which @Arh posted, ive been wanting more news on this
I ordered one during the kickstarter.
Shipment will be in August.
As said before it will never compared to a large farm of GPUs.
But if everything they say is possible, and more. This will be one heck of a device to run your own LLMs.
what was the price?
KS was $1400 USD
DAMN! I mean what can i say, you definitely need to do detailed post when it arrives
