Imboyeong/Fact-vs-Fiction-LLM-Test-

Implements multi-model comparison with automatic skill-tag-based recommendations (ideation, research, drafting, fact-checking) that route queries to GPT-4o, Gemini 2.0 Flash, or Claude 4.5 Sonnet based on task type and accuracy profiles from empirical testing. Built with React 19/Vite frontend and Express.js/PostgreSQL backend, featuring real-time side-by-side response comparison and hallucination risk warnings to help students cross-validate answers. The platform orchestrates sequential AI workflows rather than treating models interchangeably, paired with a 32-question comparative accuracy dataset (PDF report included) documenting each model's strengths and failure modes.