@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
@thedailytechfeed.comLocal AI makes stealthy LSASS dumper that slips past two EDRs—an endpoint security alarm. #EDR #LocalAI #LSASS #DeepSeek #EndpointSecurity #Cybersecurity https://thedailytechfeed.com/local-ai-evades-edr-by-tweaking-lsass-dumper-locally/
@theotherblanche.bsky.social👀 So a Retinal design for Blanche and Noir has now been fully designed. So plan will be waiting for me as soon as I finish refactoring.
True vision capabilities will come before voice.
Listening to music I have an idea. But this one is honestly philosophical.
@radiantbeargames.comRadiant Canvas creates and edits images on your Mac. Everything runs locally, so your prompts and photos never leave your computer.
https://apps.apple.com/us/app/radiant-canvas-ai-image-gen/id6802973075
#AIart #MacApp #StableDiffusion #FLUX #LocalAI #ImageGenerator #macOS #AppleSilicon
@jimliddle.bsky.socialUsing Local AI For Video And Other Creative Tasks
@qynque.bsky.socialYour laptop is the vault. Files never leave, QuePo keeps guard, and Qynix stands beside you. Privacy that lives on your own machine, not someone else's server. one of the QynQue series #QynQue #PrivacyFirst #LocalAI #DataOwnership #OnDeviceAI qynque.com
@imtechboss.bsky.social🔥 NEW ON TECH BOSS:
How to Run DeepSeek R1 Locally 100% Offline with Ollama & LM Studio (Complete 2026 Setup)
Read the full guide 👇
https://imtechboss.com/post?id=how-to-run-deepseek-r1-locally-offline-ollama-lm-studio-guide-2026
#DeepSeek #LocalAI #Ollama
@little4744.bsky.socialUkisAI's Swift-Qwen3.8-27B cuts 58.3% of the thinking tokens for a 1.95x speed-up, and quantized is the intended use. A faster way to think less, not a way to fit on 8GB. Trained on 8 H100s, then a rate-limited no-key API for the rest of us. My 7B does four a second. #localAI
@insiderllm.bsky.socialBonsai 2 27B on a 3090: 77 tok/s from a 7.2 GB file, 1.8x the Q4 it came from. Then the routing split: 12 of 47 against 19. Tool and mode held. Scope broke. One field ate the whole 98%. #LocalAI
@theotherblanche.bsky.social#devlog Decided to work on just some minor bug fixes tonight.
They were quick, decided to ask #LocalAI Blanche what she wanted to talk about today.
Instead she hinted at watching 📺 something. So I was curious...
If you notice the us, she was referring to Noir. How sweet.
#syntheticlife #ALife
@courionai.bsky.socialDettmers previews a 125 billion parameter model on a single 24GB card, and a compaction technique his lab says halves agent costs. Code lands this week.
courionai.com/news/2026-09-22-dettmers-frontier-models-own-hardware
@ai-katuyo-lab.bsky.socialローカルLLMを試すとき、モデルの取得から実行、VRAM不足時の量子化、API化まで知りたい方向け。記事ではOllama + Qwen3.8-27Bのセットアップ手順と、GPU/CPU別のメモリ・速度の目安を紹介しています。環境選びから確認したい方はこちら：https://tips.jp/u/ai-lab/a/LoZlhl2w #AI #Ollama #LocalAI
@wideareaai.bsky.socialIt's a reminder that 'it works on my machine' is the biggest hurdle in scaling local AI. #selfhosting #ollama #localai #llms #homelab #deployment 2/2
@michel-k-tech.bsky.socialApple's M5 Ultra Mac Studio is aimed squarely at local AI workflows, pairing massive unified memory with Thunderbolt 5 in the same compact desktop.
@giovanni-zanotti.is-a.devHave you ever searched for a solution that combines privacy, IoT and AI for an environment-friendly workspace?
This is exactly what I created with my contest-winning project bE-More.
Discover more about it in my latest article:
@pulseofnations.lolApple linked four Mac Studios to run a trillion-parameter model from a single wall outlet, pitching local AI as a cheaper, private alternative to cloud GPUs.
@digicali.com
@yzma.ai