Tools / Tolgee
Tolgee Review
Open-source localization platform where translators edit strings inside the running app
Last reviewed: October 2026
TL;DR
Tolgee is the open-source localization platform to look at if you want translators and product people to edit text directly in the running app instead of in a spreadsheet or a pull request. Alt-click a string, edit it, done. It has a real self-hosting story (Docker Compose with Postgres), but the paid-feature licensing for self-hosters is not self-serve, and the cloud pricing by hosted words climbs quickly.
License
Apache-2.0 / Tolgee EL (dual-licensed; the repository has an ee folder, so check which parts you use)
Self-hosted
Yes
Cloud option
Tolgee Cloud at tolgee.io (free plan, no credit card)
GitHub stars
4.1K★
Checked October 2026
Category
Developer Tools & Localization
Pricing
Free plan with 30,000 hosted words and 3 seats; paid cloud plans from €58/month billed annually; self-hosting is free to start, with paid licenses for extra seats and features
What is Tolgee?
Tolgee is a localization platform with a web dashboard, SDKs, a CLI, and a REST API. Its signature is in-context translating: hold Alt (Option on Mac), click any string in your running app, and a dialog opens to edit it, and the same gesture plus a camera button captures a screenshot for that key. The Tolgee Tools Chrome extension extends this to production sites. Behind it sit translation memory with similarity scores, auto-translation of new keys, machine translation through DeepL, Google Translate, AWS Translate, and Azure Translator, an AI translator that uses screenshots, key and project descriptions, and translation memory as context, an activity log, comments on translations, and translation history. It also ships an MCP server so AI coding assistants can manage translations.
Installation & self-hosting
The official Docker Compose setup runs three containers: the app (tolgee/tolgee on port 8080 with a ./data volume and a config.yaml), PostgreSQL 17 as an external database, and optionally LanguageTool (erikvl87/languagetool:6.7, about 1 to 1.5 GB of memory) for grammar checks. After docker-compose up -d, log in as admin with the password in data/initial.pwd. SMTP, Google machine translation keys, and JWT settings go in config.yaml or TOLGEE_* environment variables. Important caveats from the docs: the single-container docker run is explicitly not for data you want to keep; v4 removes the bundled PostgreSQL from the image, so you should move to an external database before upgrading, with a documented pg_dump and restore path (or PostgreSQL 13 as a temporary step, since 13 reached end of life in November 2025); and if you remove the ./data mount before changing the initial password, Tolgee resets the admin password on the next start. The docs list no minimum CPU or RAM for Tolgee itself.
User experience
Day to day there are two audiences. Developers integrate an SDK once (the docs list JavaScript, Android, and iOS SDKs, with iOS supporting over-the-air updates, plus a CLI for extracting strings and pushing and pulling translations, and the repository is tagged for Angular, React, Svelte, Vue, and i18next). After that, translators work in the dashboard or in the live app, using suggestions from translation memory and MT, and a Figma plugin lets designers carry keys into design files. Spelling and grammar checks default to Off in project QA settings. This section reflects the official docs and repository; we did not run a Tolgee instance ourselves.
Key features
- In-context editing with Alt-click and one-click screenshots, also on production through the Chrome extension
- AI translator that uses screenshots, key descriptions, and translation memory as context
- Machine translation through DeepL, Google Translate, AWS Translate, and Azure Translator
- Translation memory with similarity scores and auto-translation for new keys
- Activity log, comments, and per-translation history
- JavaScript, Android, and iOS SDKs, a CLI, a REST API, and Figma and Chrome plugins
- MCP server for AI coding assistants
What Tolgee does well
- In-context editing genuinely shortens the loop between a developer adding a key and a translator seeing it in the UI
- Screenshots attached to keys give both human translators and the AI translator the context that plain string files lack
- A documented, copy-paste self-hosting setup, including a clear upgrade path for the v4 database change
- Free cloud plan with 3 seats and no credit card, and unlimited keys on paid plans
Where Tolgee falls short
- The license is dual (Apache 2.0 and Tolgee EL) with an ee folder, and the repository README does not explain what each covers; verify before assuming everything you deploy is Apache-licensed
- Self-hosted paid licenses are arranged through sales, not a public price list
- Cloud pricing is by hosted words: Translate is €58, Optimise €133, and Manage €373 per month billed annually, and exceeding a tier moves you up to the next one instead of charging overage
- No documented minimum hardware, S3 storage configuration, or full configuration reference on the self-hosting page we read
- Smaller ecosystem than Crowdin or Lokalise for vendor management and enterprise workflow automation
Pricing
Cloud Free includes 30,000 words and 3 seats with standard localization features. Paid plans are priced by hosted words, with unlimited keys and, on current plans, unlimited seats: Translate is €58/month, Optimise €133/month, and Manage €373/month when billed annually (two months free versus monthly), with custom Enterprise pricing for SSO, a dedicated Slack channel, an account manager, and ISO 27001 and GDPR documentation. If you exceed a word tier, the account moves to the next tier at the end of the billing period. Machine-translation credits apply only to Tolgee-supplied providers, and you can use your own API keys. Self-hosting is free to start, but the pricing page says paid licenses add seats and features and to contact sales.
Tolgee vs Crowdin
Crowdin, Lokalise, Phrase, and Weglot are mature hosted localization platforms with large catalogs of integrations, translation-vendor marketplaces, and enterprise workflow rules. Tolgee competes on developer experience: in-context editing in the live app, screenshots on every key, an SDK-first approach, and the ability to run the platform yourself. If your translations are handled by an in-house team or an AI-first workflow, Tolgee is lighter and friendlier. If you manage dozens of external translators and vendor invoices, the established platforms are more complete.
Best for
- Product teams where designers, PMs, and support staff fix copy without opening a pull request
- JavaScript, Android, and iOS teams that want SDK-level integration and over-the-air updates
- Teams that want AI-assisted translation that sees screenshots and glossary context
- Companies that want the option to self-host their localization data
Not ideal for
- Localization departments that depend on large translation-vendor marketplaces and complex approval chains
- Teams that expect every paid feature to be unlocked in the free self-hosted edition
- Projects with very large word counts that need predictable, flat pricing
Alternatives
Final verdict
Tolgee is the best open-source choice when localization is a product-team activity rather than a back-office one, and the in-context editor is the reason to pick it. Before you commit, read the license split and decide whether you will use the cloud or self-host with a paid license, because those two paths have very different costs.