Back to docs

Configuration

Glossary and AI Model Configuration

Use glossary terms and model settings to make translations more consistent before review starts.

1 min readUpdated Mar 21, 2026

Two settings areas matter most for translation quality before review: Glossary and AI models.

Glossary

Use the glossary to define terms that should translate consistently across the project.

Each entry can include:

  • the source term
  • translated values for target locales
  • a comment
  • case-sensitivity rules

Glossary support is especially useful for:

  • product names
  • productized feature labels
  • legal phrases
  • campaign language that should not drift

Importing glossary data

If you already have terminology data elsewhere, import can help seed the project faster than rebuilding everything by hand.

Before importing at scale, verify:

  • the source locale is correct
  • locale mappings match the project
  • the import mode does what you expect

AI models

Use Settings > AI models to:

  • connect with OpenRouter
  • add or manage project AI keys
  • choose the default model
  • star preferred models for quick access
  • tune the prompt template

How to choose a model

Start with the model that gives you the best mix of quality, speed, and cost for your content type. Then compare results in the review screen before changing batch defaults.

If the team is still tuning prompts, do that on a small set of entries first. Prompt and model changes can alter tone and structure enough to affect downstream review effort.