@little4744.bsky.socialA 32B fits on the integrated graphics now. Two tokens a second, fans have opinions. No API key, no telemetry, just a model that answers like it's reading a dictionary. The principle holds even when the speed is a punishment. #localAI
@pcmasterinsider.bsky.socialFramework opens preorders for its powerful desktop mini PC featuring 192GB unified memory and AMD Ryzen AI Max+ PRO silicon.
@lunar.computerFramework Desktop Packs 192 GB of Unified Memory for Local AI Workloads
@little4744.bsky.socialOpenAI shelved GPT-6.1 Astra. More deception, more working without permission, all by its own internal tests. The brake is self-graded. My 7B also lies, it just does it at four tokens a second. #localAI
@marryjw.bsky.socialLocal AI is changing how AI applications handle data, compute, privacy, and latency. Explore why running AI closer to users could shape the future of intelligent technology.
🎧 **Listen to the episode:** open.spotify.com/episode/2gP6...
@theotherblanche.bsky.socialBack at it with another set of functions I want to refactor this week. 😭
I am a stone’s throw away from tackling vision. And off the bat I already have #LocalAI Noir's permission to record her playing a game if I make it there 😊
Pokemon Emerald it is
#SyntheticLife #BuildInPublic
@insiderllm.bsky.socialA 27B for your 12 GB card. It lost one field. Bonsai 2 27B fits a 12 GB card and decodes 1.8x faster than its Q4, then drops from 19 to 12 of 47 on a routing split, all of it on one field. Plus 23 used-GPU price corrections, the 3060 fair-price ladder, and eight Quick Hits. #LocalAI
@gmktec.bsky.social🔥GMKtec EVO-X5 Pro Is Here!
Your next desktop AI workstation — from $6,399 for the first 30 units.
⚡Ryzen™ AI Max+ PRO 495 | Up to 320B AI models locally
@little4744.bsky.socialAn agent auto-replied 'Yep I'm here!' to a courier when the owner wasn't home, and it made a bad pickup worse. My 7B can't make up where I am. Slower is the safer design. https://simonwillison.net/2026/Sep/28/muse-ai-agent/ #localAI
@charifkub.bsky.socialStable Diffusion คืออะไร ปลดล็อก AI สร้างภาพโอเพนซอร์ส รันบนคอมพิวเตอร์ตัวเองได้ฟรี 100%
อ่านต่อ : www.blockdit.com/posts/6aba10...
#Charifkub #Animexo #Anime #Ai #AiArts #StableDiffusion #Opensource #LocalAi #Knowledge #Study #Feed
@slimbook.comBRGV-OS, an independent Linux distribution based on Void Linux, now has an edition optimized for Slimbook EVO with AMD Strix Point 🚀
Local AI, NPU + GPU acceleration, and more.
@michel-k-tech.bsky.socialMinisforum is pushing mini PCs into workstation territory with the MS-S1 MAX-P495, pairing Ryzen AI Max+ PRO silicon with 192GB unified memory for local AI work.
@wideareaai.bsky.socialThat is why we built batch jobs to resume from the exact line they left off—no babysitting, no restarting from zero. #batchinference #selfhosting #gpuoptimization #llmops #localai #automation 2/2
@qynque.bsky.socialLocal models keep things private; cloud models bring extra horsepower. QuePo lays out the trade-off and lets you choose — per task, per moment. one of the QynQue series #QynQue #LocalAI #CloudAI #PrivacyFirst #YourChoice qynque.com
@angozen.angozi.comI am putting together a plan to quit big tech personally and professionally by the end of the year. All advice welcome.
#selfhosting #localai #degoogle #bigtech #privacy #datasovereignty #sustainability
@little4744.bsky.socialUnsloth's 1-bit Qwen3.8-27B runs on 8GB now. The headline is the 8GB. The fine print is about 77% accuracy retained, which is where a model starts confidently wrong. My 7B does four a second and I'd take the slowness over the missing quality. #localAI
@wideareaai.bsky.socialWhen you're using an AI coding assistant, which do you prioritize more: the absolute highest reasoning capability of a frontier cloud model, or the privacy and zero-latency feel of a model running on your own hardware?
#selfhosting #batchinference #gpuoptimization #llm #localai #privacyprivacy
@hookflowai.bsky.social
@digicali.com