---
schema: 1
type: term
title: '**style guide**'
description: >-
  A **style guide** is a set of conventions — voice, tone, formatting, and
  phrasing rules — that a model is asked to follow when producing written
  output. In this channel's material, it is used in model-comparison tests as a
  way to make writing evaluation checkable: models are assessed on how closely
  they adhere to a given style guide, not merely on whether their output sounds
  good.
canonical: 'https://liquidgold.cc/channel/everyinc/explore/term/style-guide-6e32b4f64c0f'
markdown: >-
  https://liquidgold.cc/channel/everyinc/explore/term/style-guide-6e32b4f64c0f.md
llms: null
parent: 'https://liquidgold.cc/channel/everyinc/llms.txt'
channel: everyinc
updated: '2026-10-23T22:12:35.346Z'
language: null
children: []
---


# \*\*style guide\*\*

Canonical: https://liquidgold.cc/channel/everyinc/explore/term/style-guide-6e32b4f64c0f

Parent: [Channel index](https://liquidgold.cc/channel/everyinc/llms.txt)

## Definition

A **style guide** is a set of conventions — voice, tone, formatting, and phrasing rules — that a model is asked to follow when producing written output. In this channel's material, it is used in model-comparison tests as a way to make writing evaluation checkable: models are assessed on how closely they adhere to a given style guide, not merely on whether their output sounds good.

In the channel's model-testing material, a **style guide** is a set of writing conventions — tone, voice, formatting, and phrasing rules — that a model can be asked to follow when producing text. It shows up in the context of live and week-long _vibe checks_, where writing quality is one of the capabilities being compared across flagship models: instead of just asking *is the output good*, the testing framing asks whether a model can actually adhere to a specified style rather than defaulting to its own.[^1]A style guide turns a vague writing evaluation into a **checkable one**. When you run a structured comparison of your own, giving each model the same style guide makes the outputs directly comparable — differences in adherence, not just in fluency, become part of the verdict. That's the reusable method the channel's testing videos demonstrate: hold the instructions constant across models, then judge how closely each one follows them.[^1]- A **style guide** specifies the conventions a model should follow when writing: voice, tone, formatting, and phrasing.- In this channel's material, it appears as part of live _vibe check_ testing of flagship models' writing capability.- Using one consistent style guide across models makes side-by-side writing comparisons fair and repeatable.- Adherence to a style guide is a testable signal, not just a matter of whether output 'sounds good'.