llm

3 posts.

A close view of a GPU board with its die and memory modules, the hardware whose memory capacity decides which models you can run locally.

Oct 9, 2026

Quantization: How to Run a 70B Model on the Laptop in Front of You

A 70B model is 140 GB at FP16 and about 35 GB at 4-bit — the difference between an eight-GPU rack and one consumer card. How the math works, why it barely hurts quality, and how to pick a format.

ai tech quantization llm culture guide

Rows of server racks with blinking status lights in a data center aisle, the kind of hardware that runs open-weight models at scale.

Oct 8, 2026

The Open-Weights Playbook: When to Self-Host a Model and When to Pay the API

Open 70B models now match last year's flagship APIs on many tasks at a tenth the cost — if you have the GPUs and the use case fits. A decision framework with real math.

ai tech open-source culture llm guide

An NVIDIA H100 GPU die shot, showing the tensor cores and high-bandwidth memory stacks that power modern AI training and inference.

Oct 7, 2026

NVIDIA's Nemotron 3 Ultra: The First Model Built for Agentic Workflows, Not Just Chat

NVIDIA's new flagship model isn't chasing benchmark leaderboards — it's built for the actual work AI does now: multi-step reasoning, tool use, and autonomous operation. Here's what changes when the training objective shifts from 'answer well' to 'act reliably.'

ai tech nvidia nemotron agentic-ai llm culture