Model Settings

What the sliders actually do

Model parameters change how the AI chooses the next words. They do not rewrite your scenario, repair weak memory, or make every model behave the same. They shape pacing, variation, repetition, and how adventurous the prose feels.

Start with the preset. Only adjust one setting at a time, then test for a few replies. A great setting for one model can make another model ramble, repeat itself, or become too cautious.

Understanding the Dials (Settings)

These settings work together. Think of them as steering the model's word choice and creativity, rather than simple "quality" switches. If you've never used AI before, just stick to the defaults—but when you're ready to experiment, here's what they do:

Temperature (The Creativity Dial)

Controls how bold or predictable the writing feels. Think of it like adding spice to a meal.

LowerCleaner, steadier, safer. Good for continuity, investigation scenes, and characters who must stay precise.
HigherMore dramatic, surprising, and emotional. Good for romance, chaos, dreams, horror, and messy conflict.
  • Too low can feel dry or stuck.
  • Too high can invent facts or lose track of who is in the scene.

Top P (The Vocabulary Filter)

Limits the pool of possible words before the model chooses one. Think of it like deciding how wide the AI's vocabulary should be for its next word.

LowerNarrows choices. Useful when a model gets scattered or over-decorates every sentence.
HigherLeaves more options open. Useful for expressive models that need room for voice and style.
  • Top P and Temperature both affect variety.
  • If both are high, the AI's writing might become unstable or weird.

Max Tokens (The Word Limit)

Sets how much the model is allowed to write in one reply. Think of it like setting a strict word limit for a school essay.

ShorterSnappier back-and-forth, better for dialogue, fights, and quick decisions.
LongerMore room for description, inner conflict, multi-character scenes, and cinematic pacing.
  • Long replies can stall the story if the AI just narrates around the point.
  • Short replies can feel abrupt if the scene needs atmosphere.

Presence Penalty (The Topic Changer)

Encourages the model to introduce new details instead of staying on the exact same idea. Think of it as a gentle nudge to change the subject.

LowerBetter for intimate scenes where the same emotional beat should deepen.
HigherBetter when a scene is looping and needs a new action, clue, interruption, or shift.
  • Too high can make the AI throw in random new elements.
  • Use gently for stories where continuity really matters.

Frequency Penalty (The Repetition Blocker)

Discourages repeated words and phrases. Think of it like a strict editor crossing out repeated words.

LowerPreserves verbal tics, catchphrases, and repeated emotional motifs.
HigherHelps when a model repeats sentence shapes, pet phrases, or the same description.
  • Too high can make the writing sound strained and unnatural.
  • Better for style cleanup than story direction.

Repetition Penalty (The Anti-Loop Guard)

A stronger anti-loop control used by some specific AI models. Think of it as a barrier stopping the AI from running in circles.

LowerLeaves the model's natural rhythm intact.
HigherPushes harder against loops, repeated phrases, and circular replies.
  • Some models need it. Others get worse with it.
  • Raise slowly, especially on models that write long, descriptive paragraphs.

Top K & Min P (Fine-Tuning Knobs)

Advanced filters that cap how many word options are considered or filter out very unlikely choices. If you're a beginner, you don't need to touch these at all.

LowerMore controlled and direct. Can help weaker models stay on track.
HigherMore freedom. Better for models that already follow instructions well.
  • Leave these alone unless you are experienced and a model feels noisy.

Why one model hates another model's settings

Each model has its own training style, default sampling behavior, context handling, and tendency toward repetition. Parameters amplify those traits. They do not affect every model evenly.

Expressive dialogue model

A model built for banter may work beautifully at higher Temperature, but the same setting on a lore-heavy model can make it invent family history or jump to the wrong character.

Long-context model

A model that handles huge context may need lower Temperature and fewer novelty penalties. It already has a lot to track, so pushing it to add more can scatter the scene.

Loop-prone model

A model that repeats itself may improve with Repetition Penalty or Frequency Penalty. A cleaner model can become awkward if those same penalties are too high.

Cinematic prose model

A prose-heavy model often benefits from more tokens and moderate variety. If replies become all atmosphere and no movement, reduce length or raise Presence Penalty slightly.

Roleplay tuning examples

Slow-burn romance

  • Temperature: moderate to high
  • Top P: high enough for emotional nuance
  • Presence Penalty: low to moderate
  • Max Tokens: medium or long

Investigation or mystery

  • Temperature: lower to moderate
  • Top P: moderate
  • Presence Penalty: low
  • Max Tokens: medium

Combat or chase scene

  • Temperature: moderate
  • Top P: moderate
  • Presence Penalty: moderate
  • Max Tokens: short or medium

Loop repair

  • Lower Temperature slightly
  • Raise Frequency Penalty gently
  • Raise Repetition Penalty gently
  • Tell the scene to cut to a new action

Custom Instructions

Custom Instructions live in Chat Settings and are best used for short, chat-wide rules that should stay active in the background. They work alongside your scenario, persona, lore, and model settings.

What they are good for

  • Keeping prose more restrained or more direct
  • Telling the AI not to treat hidden objects as visible
  • Stopping repeated fixations on throwaway details
  • Setting a hard preference that should apply to every reply

What they are not for

  • Rebuilding the whole scenario in a second place
  • Replacing proper lore cards or character sheets
  • Fixing a weak opening by piling on extra rules
  • Writing huge walls of instructions the model must drag through every turn
Best practice: if a fact belongs to the world, put it in the scenario, character profile, or lore system. Use Custom Instructions for behavior rules that should stay consistent across the whole chat.