EWE – a local coordination app for ensuring your model files stay in RAM
[Self-Promotion] Following the 1/10th rule. This is a promotional post for a paid tool that I released publicly today. If you are running local inference and…
Latest
Plurality Released: fully Free and Open Source AI agents/chatbot platform for local AI
Hello everyone! Some of you might recognize my user from the work I have done on Cosmos Cloud, but today I am here to talk to…
How to improve RAM offload?
I have only 12GB VRAM (RTX3060) but have enough RAM to run Qwen3.6 27B Q4 with offload. Something tells me that it won't achieve maximum performance…
Couldn’t hold back
Had been waiting for months and the cards finally got delivered today. No one at my workplace was excited, maybe because no one cares for AI…
July 2026; where are Intel’s GPU speeds today at?
Hey all, it is 1st of July, 2H26, and I hope that Intel has been catching up on their firmware support for their B50-B70 cards in…
More context window?
Hey people. I know this has been asked a billion times... but I'm a nOOb...so one more time.. I have a memory system that uses HDBSCAN…
Looking for open-source AI meeting note-takers (like Fathom, Fireflies, Notion AI)
I'm looking for open-source repositories or projects that serve as AI meeting assistants. I want a program or repository that can take notes using AI, create…
What is the biggest dense model that would fit into 128 GB RAM (at MXFP4)?
I don't care about speed, only intelligence, I am tired of giving money to Anthropic for them to just do whatever Claude 5 geopolitical interpretive dance…
What’s your actual agentic web research stack? (fully local, no cloud APIs)
Been running a fully local web research pipeline for my AI agent setup for a while now and realized I haven't seen much discussion about how…
The gap between closed and open models might be much smaller than commonly assumed, because we don’t know what closed model providers do *in addition to* model inference
When Claude dominates GLM-5.2 in benchmarks, it’s usually assumed that Anthropic has superior model architectures, superior training pipelines, and other advanced machine learning techniques that make…