Alex.bobko
Region: PT
Thursday 02 April 2026 16:09:21 GMT
126829
3363
106
340
Music
Download
Comments
NumNoo :
Wait, the next Mac mini M5 is coming. $600 for a home AI computer
2026-04-03 06:48:48
13
Time machine :
I thought my screen was cracked
2026-04-03 14:01:26
409
Dutton :
Llama 70b is like 1/100 as useful as Opus 4.6
2026-04-04 23:24:02
4
ً :
The hair
2026-04-02 16:17:46
187
dynami :
Bro.it's not even near a 40k GPU cluster, it doesn't even beat a 4090 laptop. The only impressive thing about it is how it does that much with so little power
2026-04-03 12:53:02
27
Sabi :
Llama 70B is trash, but it can run smaller Qwen models and those are 🔥
2026-04-04 01:12:25
5
Ben :
I’m sorry, but this is just blatantly false. I train ML models. I’ve been running Ollama LMs on my M4 Max Macbook Pro for years. Quantization and distillation made this possible. The unified memory architecture is excellent, but the GPUs in this SoC are nowhere near a $40k Nvidia cluster. It’s absolutely superb for consumer hardware, but your MacBook will be a furnace with any sizable context window at 70b, and you won’t get anywhere close to the frontier models in the cloud which average 1T.
2026-04-06 06:44:53
5
Paramjit Kahai, PhD :
I thought it was hair!
2026-04-04 00:41:29
14
anony_.28 :
70b is nothing … hopefully more in the near future…
2026-04-03 16:03:09
15
madharmdad :
I’m running the 70b version on my M2 MacBook Pro. Mind you it has 96G RAM and an 8TB SSD, but it runs very well. And my laptop is 3 years old at this point
2026-04-04 13:45:57
3
JTV :
Was the hair intentional lol
2026-04-05 01:33:37
4
𝙎𝙖𝙠𝙞𝙗 :
have they made M5 ultra Mac studios yet?
2026-04-03 08:46:49
7
Haeden01 :
it's true m5 pro 50% faster than rtx 5090 in ai but not games
2026-04-05 10:19:58
1
drunkenpaddy :
failed to mention that it overheats, throttles and nowhere as good as actual 4080-5090.
2026-04-04 12:38:21
1
Fureytha :
$20/month vs $8000 apple
2026-04-25 12:02:11
1
junefall :
This has to be scaled to millions of units to make sense commercially. Nvidia does not need need to worry about everything
2026-04-03 17:26:44
0
princeomar :
Google did it on a phone
2026-04-06 23:06:46
1
Jo :
Good video
2026-04-03 22:56:19
1
Momocoz :
Me thinking there is a strand of hair on my screen. 😂
2026-04-04 13:52:01
12
Zeke Mckee :
Heads up this video is all ai
2026-04-03 08:36:54
1
Walter in Vermont :
I saw this coming. Apple has been laying the foundation for this with the M series chips for a long time. Apple plays the long game.
2026-04-04 04:47:20
1
Playful Clinton :
Exactly it’s just a natural limitation of how the connection buses work
2026-04-27 03:14:01
0
Butterfly :
9 token per second so it's awfully slow for 5k laptop.
2026-05-16 16:26:28
0
610saxx_ :
claude ?
2026-05-11 19:33:32
0
Warshan :
NVIDIA saw this coming 😏
2026-06-01 20:07:29
0
To see more videos from user @alex.bobko5, please go to the Tikwm
homepage.
