Stage 9 · Public benchmarks
Hands-on comparisons, starting with five popular fields
Every comparison uses the same prompt, conditions, and published criteria. The hands-on label appears only when raw outputs and evidence are complete.
ChatGPT vs Claude vs Gemini: Korean Writing Test
Compare a source-based Korean Naver blog draft under identical editorial constraints.
Protocol published: 2026-07-26
Cursor vs Copilot vs Replit: Coding Test
Build the same small TypeScript expense tracker and run an identical test suite.
Protocol published: 2026-07-26
Midjourney vs Leonardo vs Firefly: Image Test
Create the same Korean café campaign key visual and inspect prompt fidelity and exports.
Protocol published: 2026-07-26
Vrew vs CapCut: Korean Caption Test
Transcribe and export captions from the same five-minute Korean interview clip.
Protocol published: 2026-07-26
CLOVA Note vs Notta vs Fireflies: Meeting Test
Measure Korean transcription, speaker separation, decisions, and action items.
Protocol published: 2026-07-26
Label policy
Protocol-ready pages do not invent scores. A hands-on badge requires test date, plan, raw output, processing and correction time, errors, watermark, export, cost, and evidence.