Fogbank for Code, Pinned Carriers, and a Frontier Model in 90 Gigabytes
A Sunday lineup that runs from a 2-bit quant of DeepSeek V4-Flash on a laptop, through a Java production hang most teams haven't met yet, into the most considered pushback I've read on the AI coding numbers we discussed yesterday. Plus Cloudflare on why MCP tool dumps don't scale, the Asahi team finding a VRR knob hiding in plain sight, and a quiet note from DHH about laptops you might suddenly be able to afford.
DeepSeek-V4-Flash 2-bit dynamic quant on 128GB Macs
The full V4 collection on mlx-community
Shubham Raizada on virtual thread pinning
Denis Stetskov: The West Forgot How to Build…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.