ViewTube

Novaki tests AI coding agents and models so you can see what actually works.

Claude, GPT and Codex, Gemini and others get the same tasks, the same prompts and the same starting files. Answers are scored blind, without the judge knowing which model wrote them, and every run is saved so the test can be repeated and checked.

On this channel:
• Head-to-head tests of AI models and coding agents
• What higher effort settings really buy you, and what they cost
• A weekly roundup of the AI news worth knowing, with sources

Your host is Novaki, a matchstick who reads the logs so you don't have to. The narration uses an AI voice; the tests, numbers and sources are real, and every video lists its sources in the description.

Videos are not sponsored. If that ever changes, the video will say so up front.