Momus - Local AI Lab
Region: US
Wednesday 23 September 2026 07:11:00 GMT
51591
1788
174
184
Music
Download
Comments
putundra___ :
Can you give an example of what cool things local AI can do?
2026-09-23 11:12:11
20
Figleos :
$20k just for local qwen bruh 🥀
2026-09-23 13:56:46
56
HarshReality :
your room sound like inside of an airplane
2026-09-23 08:19:07
247
El ricardio :
20$ subscription can do it better
2026-09-23 20:36:08
12
LordPiksy :
Crazy how this loses in speed and output to a 20$ subscription
2026-09-23 14:43:05
16
pingpong :
a lot of heat and a lot of power consumption, tell Nvidia to start creating non von Newman calculator chips. This is an architecture of 1945 it is so slow.
2026-09-24 20:30:26
0
Chiel.G :
What are you actually using this for?
2026-09-23 10:01:38
26
Rob Hartman :
Your room must be 120 degrees.
2026-09-24 21:52:40
0
MetalDrgn :
why would you buy 5090s instead of a 6000?
2026-09-23 17:07:19
2
emretoilet :
Just larping atp
2026-09-23 12:10:10
7
Voll :
Not a smart choice unless you wanna run uncensored models (even then there are cheap providers). Assuming youre ONLY running the better GLM 5.3 via API, 20k would get you 10 Million tokens a day for 3 entire years. Your hardware will be outdated by then. To generate 10 million tokens a day you‘d also be using enough electricity to bring the break even point further down.
2026-09-23 07:34:59
3
nixusKnight :
$20 a month is way more efficient than dropping $20k.
2026-09-24 21:59:26
0
mnkyooby :
Ngl that’s mad funny had Astra make me a vfx for a Roblox game that looks like that 😂
2026-09-24 03:10:29
1
E I R E N O X :
why not using exl3 instead?
2026-09-24 06:54:31
1
ابو سالم :
i run qwen on my vega 56 😂😂😂
2026-09-24 04:45:51
0
Ali Youssef :
All u need now is jev
2026-09-24 01:52:36
0
Giro :
what do you use to load the llms? or to see the tokens per second?
2026-09-23 07:40:22
0
Dash :
et tout ça, pour faire quoi au juste ? 🙂
2026-09-24 05:48:50
0
smurfbox :
2026-09-23 10:12:36
6
Cheliotop :
Just read the text and imidiatly subbed
2026-09-23 13:41:49
2
https://kastal.ing :
why use 3 5090s for qwen 3.8 27b. wouldn't 2 4090s be enough even? or are you going for speed/max context
2026-09-23 07:15:40
1
nicafloki :
You need 100 months of 200usd claude for spent 20k 😂😂😂😂 in 8 years you need to change that 20k hardware 😂😂😂
2026-09-24 15:15:51
1
A..Kak..Kakatb :
It might reach the Sonnet level
2026-09-24 04:50:12
0
Ryan Arthur :
Now we’re cooking
2026-09-23 16:18:50
1
Gorilla007 :
10 Tokens er hour. yeah we know
2026-09-23 15:05:49
3
To see more videos from user @momus.ai, please go to the Tikwm
homepage.
