14:30 Benchmarks2 months ago MiniCPM5-2B: The Best Sub-Agent Model Yet? Sam Witteveen examines MiniCPM 2B (technically 2.5B), the latest small model from OpenBMB, which claims to beat 4B-class models on fu... 0 comments 4.4K views
38:46 Tutorials3 months ago Local AI Is Better Than You Think (Gemma, HuggingFace etc) Greg Isenberg delivers a comprehensive introduction to local AI for non-technical founders, arguing that running open-source models o... 0 comments 1.9K views
08:53 Reviews & Comparisons3 months ago MiniCPM5-2B GGUF: Runs Great, Until It Doesn’t Fahd Mirza puts the MiniCPM 5-2B GGUF model through a structured battery of tests on an Ubuntu system using llama.cpp, focusing on th... 0 comments 2.5K views
15:27 Tutorials3 months ago Qwen3.8-27B GSQ+RCO: 27B Params, 11.8GB, Zero Accuracy Lost Locally Fahd Mirza covers two new quantization techniques — GSQ (Gumbel-Softmax Quantization) and RCO (Riemannian Constrained Optimization) —... 0 comments 2.9K views
10:19 Coding & Builds3 months ago OmegaClaw: An AI Agent Built on Symbolic Logic, Not Just an LLM OmegaClaw is an open-source agent framework from SingularityNET that takes a structurally different approach from most LLM-based agen... 0 comments 1.9K views
16:41 Tutorials3 months ago Run GLM-5.3 Flash Locally on CPU + RAM Fahd Mirza walks through a complete, reproducible tutorial for running GLM-5.3 Flash — ZAI's 320-billion-parameter sparse mixture-of-... 0 comments 8.6K views
11:07 Reviews & Comparisons3 months ago Qwen3.8-Flash-Next: Local Install, Serve, Break, Fix Fahd Mirza takes Qwen3.8-Flash-Next for a hands-on spin the day it drops — an open-weight preview of the upcoming Qwen 4 architecture... 0 comments 3.4K views
08:43 Deep Dives3 months ago Qwen3.8-27B Obliterated: Don’t Use this Model in Production Security-focused AI researcher Fahd Mirza examines "obliterated" open-weight models — specifically Qwen 3.8-27B Obliterated, availabl... 0 comments 4.3K views
11:12 Benchmarks3 months ago Qwen3.8-4B Distilled: Q4 vs Q6 vs Q8 — Which Quant Actually Wins? Fahad Mirza benchmarks the Qwen 3.8 4B distilled model — built by the Emperor team, not Alibaba — across three quantization levels (Q... 0 comments 3K views
09:20 Benchmarks3 months ago DFlash 2: Qwen3.8-27B at 2× Speed – Live Benchmark Locally Fahd Mirza benchmarks DFlash 2, a community-developed speculative decoding enhancement for the Qwen3.8-27B model, running live on an... 0 comments 6.6K views