Docs AI Translation

AI Translation

AI-powered translation with variable protection, brand voice consistency, and smart caching.

Overview

i1n translates your app to any language using AI. Push your keys, specify target languages, and translations are ready in seconds — with full variable protection and brand voice consistency.

Translations are processed in parallel with support for 200,000+ strings simultaneously. Common translations are cached across the platform, so repeated strings translate instantly at reduced cost.

Variable Protection

Interpolation variables like {name}, {{count}}, and %{user} are automatically protected during translation. They are never modified, reordered, or corrupted — regardless of target language or script direction.

This works out of the box with all supported formats (JSON, YAML, ARB, XML, .strings, TypeScript). No configuration needed.

Brand Voice

Define your brand's tone once in the dashboard — formal, friendly, technical, or concise. Every AI translation follows it consistently across all languages.

Available on Pro plans and above. For domain-specific wording, Business and Enterprise plans add Terminology rules: per-language term rules the AI enforces on every translation.

Translation Cache

Common translations are cached and reused. When a string has already been translated (by you or by the community on the public cache), it resolves instantly at a fraction of the cost.

Cache tiers by plan: Starter uses a shared public cache. Pro uses mixed (private + public). Business uses private-first. Enterprise gets a fully dedicated, isolated cache.

Cache hit costs vary by plan: 0.2 WU (Starter), 0.1 WU (Pro), 0.05 WU (Business), 0.01 WU (Enterprise).

Credits

Each AI translation costs 1 credit. Cache hits cost less depending on your plan (from 0.2 down to 0.01 credits). Credit allowances reset monthly.

Example: translating 100 keys to 3 languages costs 300 credits if none are cached. With a populated cache, costs drop significantly. Monitor usage with i1n limits.

Monitor your WU usage with the i1n limits command.

Retranslate

Retranslate re-runs every AI translation of one language with your current tone and terminology rules. It never uses the cache, so the output always reflects the rules you have configured today.

Hover a language in the dashboard sidebar and click the refresh icon. Before running anything, a dialog shows how many values will be retranslated and the exact cost. Values your team verified or wrote by hand are kept by default; an explicit checkbox lets admins replace them too, and everything it replaces goes back to review.

Available on Business and Enterprise plans, for owners and admins. While a retranslation is running, other translations for the project wait their turn, so nothing gets overwritten halfway.

Related