@maxveroo: Calibri just ran a 744 billion parameter AI model on a laptop without a GPU, using smart RAM and hard drive streaming. This breakthrough challenges NVIDIA’s GPU dominance and offers an accessible offline AI experience for developers and tech enthusiasts. #AI #AITools #TechInnovation #MachineLearning #OpenSourceAI
1 token per sometimes besides doesn’t that mean a normal 10 gb model will do just as good? Cuz only 10 gigs are loaded
2026-10-01 08:35:02
20
adri_.sw :
1 token per blue moon
2026-10-01 14:46:55
13
Dave 🌙☀️👑 :
just math but not efficient why takes lot to run models
2026-10-01 17:22:58
0
user4555059709567 :
it's been a week 1 token so far 😂
2026-10-01 20:11:35
2
Mufaro :
1 token per sometime
2026-10-01 10:52:16
1
Stefano olsen :
1 tokens/Tuesday
2026-10-01 12:36:19
31
Mr.Hunor :
1 token / universe birth
2026-10-01 15:59:00
11
Liggma :
.1token per decade
2026-10-01 16:56:02
0
®𐩨𐩡𐩺™ :
token in a minute😂
ᅠ ᅠ
I am a young man in my twenties, and my grandchildren will inherit my project, which will reach its conclusion in their time.
2026-10-01 14:47:10
8
MR. N... 🌃✨ :
I know it will be extremely slow on this laptop... but imagine it on a gaming pc... it could be interesting
2026-10-01 17:58:59
0
Delusional :
O.32 tokens per sec on m5 pro with 48gb after two days of downloading.
2026-10-01 19:29:34
1
jelly :
Ram crisis could finally end
2026-10-01 01:22:26
4
some one :
ah jes, no gpu...
so he does things on a blackscreen?
2026-10-01 20:15:17
0
MrTuk :
But ram is extremely slow, you might have to wait minutes for easy prompts or even hours for more complex ones. As soon as the vram is full and even one bit is offset into the ram, the speed dies down insanely fast.
2026-10-01 10:53:01
0
Maria mims :
1 token
2026-10-01 21:48:37
1
Twisty :
1 token per sometimes
2026-10-01 16:37:06
4
To see more videos from user @maxveroo, please go to the Tikwm
homepage.