@hackswithbanks4: 🚀 Raspberry Pi just got a serious AI upgrade! The new Raspberry Pi AI HAT+ 2 adds a 40 TOPS Hailo-10H AI accelerator and 8 GB of onboard memory, turning a Raspberry Pi 5 into a much more capable local-AI machine. 🤯 💻 What can you use it for? • Run supported AI models locally • Object detection & computer vision • Robotics and smart-camera projects • Local LLM experiments • Vision-language applications • AI projects without sending everything to the cloud 🛠️ Basic setup: 1. Get a Raspberry Pi 5 + AI HAT+ 2 2. Connect the HAT through the Pi 5's PCIe interface 3. Install Raspberry Pi OS and the required AI software 4. Load a supported model 5. Start running AI inference locally The big idea: move more AI processing to the device instead of relying entirely on cloud servers. 🔥 ⚠️ Educational/technical content. Always follow the manufacturer's documentation when assembling hardware. #RaspberryPi #AIHAT #RaspberryPi5 #ArtificialIntelligence #EdgeAI
I can use old phone to run bonsai , why spend so much money for pi?
2026-09-22 21:36:28
10
harvestofshadows :
By the time you drop $300–$400 on the Pi setup and $200 on the AI module, you're in mini PC territory with way better performance.
2026-09-21 14:26:40
1257
Stokamemes :
1 token per sometimes
2026-09-22 15:23:01
590
Ragnarr Loðbrók :
2 token / sec
2026-09-21 19:12:29
312
7empest :
Qwen2 1.5b 😂
2026-09-21 15:59:15
136
Poleznyy :
Qwen 0.01b
2026-09-21 22:27:20
320
Varghese Muthalaly :
How would you compare it to the Nvidia Jetson
2026-09-23 05:22:11
9
codegrad :
What are some use cases for a model that small in this mobile contraption?
2026-09-21 19:32:53
13
Ragnar Lothbrok :
£200 to run a 1.5B model? 😂
2026-09-21 22:45:42
112
devolt :
8 GB RAM, yeah you can run some local AI fart generator
2026-10-06 07:33:57
38
exean :
This should be illegal.It doesn’t even look like a hat 🎩
2026-09-21 19:51:11
96
Just Some Guy :
For playing/ testing this looks fine, but for real work, memory is going to be an issue. I can't see LLMs working properly on this
2026-09-21 14:00:19
9
Tornd :
Does the price will get lower ? Cause 200€ is quite a lot 😅
2026-09-21 20:47:59
10
AhhhFiddleSticks :
Can someone tell the tech influencers that we can't afford Raspberry Pis anymore
2026-09-26 21:19:41
16
انه لجهاد نصر او استشهاد :
can we cluster multiple for big models
2026-09-23 06:27:51
0
LDMEC :
@LDMEC:@LDMEC:my hat 1 20tops no ram runs my outside camera rtsp feed runs full object yolo8n put all my inference code in the g stem pipe , dont read frames , still get preview tho , 12percent CPU only and runs 247 😎✌ 69 degrees
2026-11-16 03:58:58
0
Soibo1 :
local ai and open source is the good version of the future
2026-09-23 21:52:35
12
F3ZY :
8gb ram? shi must be expensive
2026-09-24 14:21:53
14
FatPurple :
I currently run a 3b model on my Pi 5 without the HAT. Having a dedicated extension would be nice though 🔥🔥
2026-09-22 13:20:05
7
Daniel Moreno :
this is for?
2026-11-14 08:43:24
0
yeitsme :
if you want to sell this to me, explain real life usage scenario. because atp id rather connect to claude via work account at home as well
2026-10-19 17:39:35
0
I_love_the_kites :
Who is running a model on a PI? [Laugh approved]
2026-09-24 20:29:02
0
Grim The Chaos Jester⚓️🏴☠️🃏 :
The question is, is, like, are you able to get what parameter models uncensored, unethical, and also are they able to do analysis or inference generation?
2026-09-26 06:34:53
1
dip_saus :
Is 8 gig enough? I mean what can you do with less than a 8 gig model? In terms of context? How much output before it starts to hallucinate??
2026-09-21 16:53:36
2
ryankylecole :
What AI models are you running with 8 gigs that have an use cases..
2026-09-23 23:08:22
4
To see more videos from user @hackswithbanks4, please go to the Tikwm
homepage.