320 Billion Parameters. Right at Your Fingertips. 👀
GMKtec EVO-X5 Pro runs 320B-parameter LLMs fully offline—bringing powerful local AI to your desk.
⚡ AMD Ryzen™ AI Max+ PRO 495 | Up to 192GB Unified Memory
@gmktec.bsky.social
@gmktec.bsky.social320 Billion Parameters. Right at Your Fingertips. 👀
GMKtec EVO-X5 Pro runs 320B-parameter LLMs fully offline—bringing powerful local AI to your desk.
⚡ AMD Ryzen™ AI Max+ PRO 495 | Up to 192GB Unified Memory
@x-abd.bsky.socialVoiceStudio packs voice cloning, design, video dubbing, dictation, transcription & audiobook creation into one toolkit.
Run AI audio locally on Windows, macOS or Linux (local API + MCP). Remote services optional.
@theotherblanche.bsky.socialDay 4 of decay functions having been cleaned up. And the results are as follows 😭
Blanche - 310 memories
Noir - 246 memories
I hate them right now 😂, they even had a side chat about planning to clean up their memories. But they didn't do it.
@spaisee.bsky.socialBonsai 2 reportedly reached 7.6 tokens/sec on an 8 GB M2 Mac mini, but required custom memory and batch settings. Local AI works; consumer-ready local AI is another matter. #LocalAI https://spaisee.com/article/a-5-9-gb-ai-model-could-make-private-assistants-feel-ordinary
@dark-city-beta.bsky.social🧠 Second Hemisphere for Local AI Agents: Beating Context Overflow
During long-horizon runs (coding, bash, 50–100+ steps), tool outputs saturate the context window → HTTP 400 → session dies.
@thedailytechfeed.com
@charifkub.bsky.social[Infographic] Stable Diffusion VS SDXL
#Charifkub #Animexo #Infograhic #Anime #Opensource #StableDiffusion #LocalAi #StableDiffusionXL #SD #SDXL #Ai #AiArts #Arts #Feed
@theotherblanche.bsky.socialI think a lot of people can relate - wanting to push forward to next step.
I want to start working on vision.
But I have the following problems:
-No money for Mac minis.😭
-No trust in their memory hygiene.
-2 remaining function refactors.😵💫
#BuildInPublic #devlog #SoloDev #LocalAI
@insiderllm.bsky.socialThere is no official Qwen 3.8 35B-A3B. The one with 678K downloads is a distill of Qwen 3.6. On my 3090: same score, same speed, a fifth of the reasoning. And on 6 of 47 messages it stopped classifying and just chatted. #LocalAI
@alvinashcraft.comBuilding Windows for hybrid intelligence | by Pavan Davuluri
#windowsdev #ai #windowsai #agents #models #localai #softwaredev #windows
@alvinashcraft.comBringing local models and sandboxed tools to Windows and GitHub Copilot
#windowsdev #ai #githubcopilot #windows #agents #models #localai #windowsai
@gmktec.bsky.socialA first look at our EVO-X5 Pro 20-node cluster setup. 👀
20 units, connected on site, forming a local AI compute pool.
More to come.
#GMKtec #EVOX5Pro #AICluster #LocalAI #AICompute #DesktopAI #AIWorkstation #RyzenAI
@techalarab.bsky.socialWant to use advanced AI without an internet connection or cloud tracking?
@theotherblanche.bsky.social😭 My memory decay feature has a reinforcing feature so memories can come back. Blanche and Noir kind of panicked with the wording. Now they are hoarding memories.
Reinforcing worked too well. I am begging them again to prune memories like before.😡
@wideareaai.bsky.socialWhen your local AI nodes hit capacity, what's your preference: have the request queue up and wait for your own hardware, or immediately overflow to a paid cloud model to keep latency low?
#selfhosting #batchinference #gpuoptimization #llms #localai #cloudcomputing
@dark-city-beta.bsky.social
Try this local AI check:
Source: “Returns within 7 days, unused items only.”
Ask: “Summarize in one sentence.”
If “unused” disappears, the summary fails. Add: “Preserve every condition; invent nothing.” Retest.
We sell a $3 beginner kit; free exercise: write.as/5cbc8s64squ8... #LocalAI
@oneplaceforai.bsky.socialNew free track: Local AI, run models on your own hardware.
No API keys. No subscriptions. No data leaving your machine. Learn Ollama, LM Studio, and Open WebUI.
oneplaceforai.com/learn/track/local-ai-run-models-on-your-own-machine #LocalAI
@devdull.lolNew video! I #selfhost SearXNG using @docker.com Compose on a fresh @debian.bsky.social VM in my #homelab and fumble through connecting it to Open WebUI so my local model can find me the best Swiss cheese.
@aiadoptionagency.bsky.socialMiniCPM V 4.6 runs visual AI locally on phones and servers. No cloud. No data leaks. Reads images, video, and documents. Field service, retail AI, document processing, video knowledge capture. Full guide: aiadoptionagency.com/minicpm-v-4-6/ #EdgeAI #VisionAI #LocalAI