Test 1: Janitor LLM (Default Model)

withVance

Tested using Janitor AI's native default LLM. Shows a surprisingly good, natural balance between in-character dialogue and structured code formatting without needing a proxy.

@razor0814·2 months·4