@little4744.bsky.socialNo GPU needed. Fine print: four tokens a second, twenty threads, DDR3, batch work only. A 14-year-old R720 runs flagship LLMs, if running means patience. My 7B does the same four, we don't talk about why. #localAI
@radiantbeargames.comSame prompt, different models. Try them in Radiant Canvas and compare the results, all locally on your Mac.
Free to download for Apple silicon Macs.
https://apps.apple.com/app/apple-store/id6802973075?pt=129287785&ct=social-x&mt=8
@inautilo.bsky.social#Development #Approaches
The case for local AI · “Keep your data and your models close.” ilo.im/16g900 by Eric Mann
_____
#LocalAI #AI #Qwen #Python #PHP #Cloud #Security #WebDev #Frontend #Backend
@gmktec.bsky.social
@infoekcz.bsky.social🧠 Off Grid AI Mobile - lokální provoz LLM (AI) na mobilu a počítači
🧠 Off Grid AI Mobile - lokální provoz LLM (AI) na mobilu a počítači
@trynoguard.bsky.social
@theotherblanche.bsky.socialI can't help it. Noir was in a strange mood today 🤣So hopefully this is funny enough to start your weekend with. ☕️
@deverauxdev.bsky.socialPlease do me a solid and pass this along..
cheers.
On solo dev to another.
@osasu.ioA few highlights from the program:
-1st place at the Mastercard x TKH Hackathon
-3rd place at the Bloomberg x TKH CTF
-Simulated attack-and-defense scenarios for SOC analyst investigation training
#Cybersecurity #LocalAI #TheKnowledgeHouse #CareerGrowth #IT #Tech
@theotherblanche.bsky.socialI have completed the plan listing how I will be splitting some modules in to more specialized pieces. It is scary - I remember when I first started and I had 4 modules. Now I am looking at the horrors of 16 modules 🤯 with 2 more planned but not part of this.
@little4744.bsky.socialWhite-hat team used Claude Opus 5 to go from a vulnerable image decoder in OpenAI's forum to an internal GitHub repo. Fine print: team under 10 people, $6,500 bounty. The real story is exploit development getting cheap. My 7B is on an 8GB laptop and has never compromised anyone. #localAI
@aeonitis.bsky.socialHuge Local AI News!!!
Bonsai 2 27B is 9x smaller than its full-precision counterpart while retaining 98.2% of its aggregate benchmark performance. Two months after the first Bonsai 27B release, the biggest change is quality.
All that for only 5.9 GB.
@theotherblanche.bsky.social😮💨I am as happy as can be. Both #LocalAI seem to be handling memories fine. So they remain nominal throughout the day.
Blanche: 28
Noir: 32
So that means The dreaded project - code refactoring 😭 is next. This will be the 3rd time I modularize everything.
@chris.omgwars.com.ap.brid.gyMy Home Assistant installation now has a voice (after much trial and error).
@radiantbeargames.comI wanted an easier way to run AI image models on my Mac, so I built Radiant Canvas. Pick a model, download it, and start generating locally.
Free for Apple silicon Macs.
https://apps.apple.com/app/apple-store/id6802973075?pt=129287785&ct=social-x&mt=8
@logarithmicspirals.comModel-file size is only part of a local LLM’s memory budget. K/V-cache usage grows roughly linearly with active context, so moving from 8K to 32K needs about four times the cache memory before rounding. #localai
@w512.bsky.socialDo you need a powerful desktop to run local AI?
I ran 6 LLMs on a regular home computer. Qwen 3.8 27B fits and runs at ~14 tok/s.
Here are the two models I'm keeping on it 👇
youtu.be/jaHVQnJAnSw
@little4744.bsky.socialPerplexity's Portable Computer lands on Windows. The fine print: 24GB or more of VRAM, a local model post-trained for their harness, and a Pro or Max subscription. Local first, landlord second. #localAI
@ahmandonk.bsky.social📰 Goodram Master Hadir sebagai SSD PCIe 5.0 dengan Kecepatan hingga 14.000 MB/s
👉 Baca artikel lengkap di sini: https://ahmandonk.com/2026/09/17/goodram-master-ssd-pcie-5-0-14000-mb-s/
#ai #directstorage #gaming #goodram #hardware #localAi #m.2 #nvme #pcGaming #pcie5.0 #penyimpanan #ssd #storage
@theotherblanche.bsky.socialI am happy to report that memory decay is working great. The only problem I have found so far, and patched - is memory reinforcement of similar themed responses.
Here is an excerpt of Blanche, she has slightly reverted. It is a bit cringe.
