Localize.io

A supplied style guide is applied to translation jobs and can affect register choices alongside memory and other context.

Customers can add their own people as account or project reviewers. They sign in to assigned work and approve or correct each string, with verdicts and notes returned in review-job results.

Provides REST API and MCP integration; orders and results can be exchanged as API-schema data, files or plain text, and Enterprise describes custom connectors.

Customers can choose listed model providers and configure model routes per language, content type, project and pass; tour examples show different routes for UI, marketing and review work.

Supports customer provider keys and Enterprise instances on customer infrastructure with customer model accounts.

Uploaded domain glossaries and style rules condition model translation output; per-order instructions provide a brief for the submitted content.

Returns translated files or structured results in the submitted shape, checking identifiers, placeholders and tags for downstream integration.

Glossary terms are selected for relevant strings; automated terminology checking identifies failures and sends them through retry before delivery.

Imports TMX/XLIFF translation memories and TBX terminology as reusable linguistic assets, and lets customers download their original uploaded files unchanged.

A customer can set a monthly USD budget cap; the budget endpoint reports delivered spend, estimates in-progress work and refuses an order that would exceed the cap.

Accepts screenshots and per-string context along with glossary, translation memory and style rules; the tour shows that these inputs are matched to strings to give translators context.

Job traces retain routing, model calls, supplied context, checks, warnings and costs; account logs cover job and asset events.

Stores customer- and project-level memory assets, keeps asset versions, and permits restoring a prior file.

Stores multilingual approved-term files at account and project levels, allows replacement/removal and asset-version restoration, and applies project terms over conflicting account terms.

Routes work to configured models by language, content type, project and pass, and can name a human reviewer for a review pass.

Customers can buy a human review of translated content through the API using a service-provided linguist listed with language pairs, content types and per-character pricing.

Submits strings or files for AI translation; configured or order-selected language models translate them with applicable glossary, memory, style and screenshot context.

Lists delivered-job character counts and prices, with totals by project and API key and monthly cost views.

Each string is given exact and semantically similar translation-memory matches; the result identifies whether a remembered translation was kept, corrected or newly written.

Runs automated checks for returned strings, placeholders, tags, length limits, line breaks and terminology, with failed strings retried and remaining failures flagged.

The documented API workflow has a translation job followed by a separate review job naming a person; the reviewer approves or corrects each string and automated checks run again.

Onboards the customer’s own reviewers by name, email, language pairs, rate and optional project; reviewers sign in and see their assigned review work.