| LLM | Large Language Model. The AI that powers conversations. Different LLMs produce different quality responses. When you select a model in chat settings, you're choosing which LLM runs your character. |
| Tokens | Pieces of words. Roughly 1,000 tokens equals about 750 words. Everything you write in your character's profile uses tokens, and so does every message in the conversation. The more tokens your description uses, the less room the AI has for actual conversation memory. |
| Temperature | How creative or unpredictable the AI is. Lower temperature (0.0 — 0.5) is more consistent and logical. Higher (1.0 — 2.0) is more creative but also more random. Default is 0.8. |
| Character Card | The full profile that defines who your character is. Built from what you write in the description, personality, and backstory fields. |
| Chat Mode | Your character texts you like a real person. Short messages, slang, emoji, typos — like iMessage. No physical actions. |
| Roleplay Mode | Your character is physically present in a scene with you. They perform actions, show inner thoughts, and describe surroundings. Closer to collaborative fiction. |
| Scene | A pre-built scenario for roleplay. You define location, situation, and mood. The character adapts to the scene while keeping their personality. |
| Persona | A description of who YOU are to the character — your name, age, relationship. Without one, the character discovers you through conversation. |
| Trust Level | A number (1 — 6) tracking how close your relationship has become. Starts at Level 1 (Stranger) and deepens over time. Full details in Chapter 19. |
| First Message | The opening message your character sends when a new conversation starts. Also called the greeting. Brief in chat, immersive in roleplay. |
| {{char}} | A placeholder that is replaced with your character's name. If your character is Elena, {{char}} becomes Elena. |
| {{user}} | A placeholder that is replaced with the user's name (or their persona name). |
| Max Tokens | A per-chat setting (64 — 5,000) controlling how long the AI's responses can be. Default is 200. |