If LLMs Can Decide Without Fine-Tuning, Do We Still Need Models Like Jev? (itsodeleo.github.io)