Local Models, Real GPUs

Run AI models on real hardware. See what actually happens.

A YouTube series testing AI models on real GPUs. Same prompts, same rigs, one variable changed per test. All results published.

Methodology

Same Prompt

Every test uses identical prompts to ensure fair comparison between models.

Same Rig

Consistent hardware setup across all tests for reliable, repeatable results.

One Variable

Only the model changes between tests. Everything else stays constant.

Model Reloaded

Each model is freshly loaded before every run for accurate performance.

Results Published

Every prompt and result file is made publicly available for verification.

Transparent

Full transparency in methodology ensures trust in the results.

Hardware

Test Rig

Every test runs on the same hardware setup to ensure consistency and repeatability. The results you see are from real, local execution on actual GPU hardware.

RTX 3090
GPU
24 GB
VRAM
LM Studio
Server
Local
Execution
NVIDIA RTX 3090
GA102
Ampere Architecture

About PC Magik

PC Magik is a YouTube series dedicated to testing AI models on real hardware. The goal is simple: show what actually happens when you run AI models locally.

Created by Mateusz Piekut, a computer service and homelab operator from Poland. The series focuses on transparency, consistency, and real-world performance.

Every test follows strict methodology: same prompts, same rig, one variable changed per test. All results are published for verification.

MP