[ Video ] Running Nanbeige 4.2 3B on my 16GB MacBook Pro
I tested Nanbeige 4.2 3b on a 16GB MacBook Pro. The 4bit quant is good, but the generation is relatively slow.
[ Video ] Running K2 Horizon 3.7B on my 16GB MacBook Pro
K2 Horizon 3.7B runs comfortably on a 16GB MacBook Pro. Worth noting that the 7B parameter version was better, but way too slow.
Can AI Improve Itself Without Training? Well, Maybe…
A frontier model was capable to change a smaller model by just editing their weights. The short and long term consequences are fascinating.
Running a 1.7 Billion Parameter Model – LFM3 1.7B – on a Raspberry Pi 4
A Raspberry Pi 4 can run comfortably (3t/s) LFM3 1.7B. Not usable for heavy coding, but good enough for an orchestrator.
[ Video ] An AI Model in your Browser? LFM3 Omni 2.2 on my 16GB MacBook Pro
I tested Liquid Foundation LFM3 Omni 2.2B model, in a browser on my MacBook Pro. At 1.6GB download, the experience is spectacular.
[ Video ] Testing Ling 3.0 Tiny on my 16GB MacBook Pro
Ling 3.0 Tiny is a surprisingly good model to run on a 16GB MacBook Pro. Test generation can go as fast as 45 tokens / second.
[ Video ] Testing Apple Intelligence on my 16GB MacBook Pro
I tested the Apple Intelligence model Apple ships with its default install, by actually chatting with it. One of the fastest models so far.
[ Video ] Testing Spark-X2.6-4B on my 16GB MacBookPro
Spark 2.6 4B from Spark LLM is a decent model for its size. In my testing, there was even room left for other apps on my [...]
[ Video ] Testing Nemotron 3.5 Bolt E4B 4GB on my 16 GB MacBookPro
Testing Nemotron 3.5 Bolt E4B 4GB on my 16GB MackBook Pro: balanced answers and decent speed. It's a good model for day to day usage.
[ Video ] Stories From The Vibe Coding Trenches – When the Agent (Claude) Deletes your Files
During development of one of my apps, Zen Tales, my coding agent (Claude, back then) deleted some audio files I already paid for. A difficult lesson.
[ Video ] Stories From The Vibe Coding Trenches – The User Knows Better Than You
Second installment of my stories from the vibe coding trenches is about a mindset shift: nothing is difficult anymore. Listen to your users.
[ Video ] Running Nemotron Dory 1.0 4B on my 16 GB MacBookPro
Nvidia Nemotron Dory 1.0 4B is a surprisingly good model that can be run locally, on a 16GB MacBook Pro, for coding and text generation.
[ Video ] Stories from the Vibe Coding Trenches – Accidentally Overpaid $150 to Cloudflare
This are real stories from my life as a vibe coder. This one shows how I overpaid $150 on Cloudeflare while vibe coding Flight Lens.
[ Video ] Running Liquid AI LFM V2.5 on my 16 GB MacBookPro
The 4th video in the series about what models can you realistically run on a 16GB MacoBook tests Liquid AI LFM 2.5. A solid model.
[ Video ] How To Set Up Local AI on a 16GB MacBook
Setting up local AI on a 16GB MacBook involves using optimized binaries for the Metal architecture. Start with MLX-serve. Optionally OpenWebUI
What Makes an AI Model a “Really Good” AI Model?
At the end of the day, all AI models are the same: statistical approximation machines. A really good AI model is just in the question you [...]
[ Video ] Running Xiaomi MiMo V2.6 (Qwen 9B distill) on my 16 GB MacBookPro
This week I'm testing Xiaomi MiMo V2.6, Qwen 9b distill, on a 16GB M1 MacBookPro. It works good enough for text generation and light coding
[Video] Running Google Gemma 4 E4B on my 16GB MacBookPro
This week I'm testing Google Gemma 4 E4B on a 16GB M1 MacBookPro. It works surprisingly well for text generation and even light coding.
[Video] I Tested Grok 4.7 With One Simple Job: Turn My vi Task Files Into a Bash Agenda
Grok 4.7 was launched yesterday and I had a chance to test it. It has better prompt adherence, it's more thorough, but also token greedy.
[ Video ] A System One AI Doesn’t Talk, It Only Decides – And It’s Perfect for Assess Decide Do
A System One model, like JEV, doesn't output text, it only decides, and that's why is perfect for my Assess Decide Do framework.