PC Magik Lab

Local models, real GPUs.

About

Welcome to PC Magik Lab, a dedicated exploration of the cutting edge of local AI. This series focuses on the practical performance of Large Language Models (LLMs) when run on enthusiast-grade hardware, providing transparent and reproducible benchmarks for the community.

Lab Standards

Precision Testing

  • Same prompt for every test
  • Same hardware rig throughout
  • One variable changed per test

Strict Isolation

  • Model reloaded before every run
  • Clean state for every measurement
  • Zero interference from other apps

Full Transparency

  • Every prompt published
  • Every result file published
  • Open for community verification

The Rig

GPU

NVIDIA RTX 3090 24 GB

Model Server

LM Studio