Originally published at devtoolpicks.com
Nothing about localization pricing feels expensive when you're testing a demo with two languages. It starts to bite the moment your product actually works: a third language gets requested, a translator who isn't a developer needs to touch strings, and suddenly you're comparing per-word rates against your MRR.
Four tools cover most of what an indie hacker actually needs here, and the honest finding this year is that the two most recognizable names moved in opposite directions on price. One got sharply more expensive. One quietly kept its free tier intact.
Quick verdict
| Tool | Best for | Price | Rating |
|---|---|---|---|
| Crowdin | Lowest-cost hosted option with a real free tier | Free, then $59/mo | 4.5/5 |
| Lokalise | Teams that need AI-scored translation quality | No free plan, $144/mo | 3/5 |
| Tolgee | Self-hosting on your own budget | Free self-hosted, or cloud from ~$42/mo | 4/5 |
| Weblate | Full open source, zero license cost | Free self-hosted, hosted from ~€19/mo | 4/5 |
Is Crowdin still the safe default?
Crowdin has been the developer-facing default in this category for over a decade, and the pricing still reflects that positioning. The free plan covers open source and personal projects with one private project and 60,000 hosted words, where hosted words means source words multiplied by target languages. Upload 500 words with 10 target languages and you've used 5,000 of your allowance.
Paid plans start at $59/month for Pro, then $179/month for Team, and $450/month for the Team+ tier, with a custom Business plan above that. No per-seat charges on any tier, translators and proofreaders are unlimited on paid plans, which matters if you're coordinating unpaid community translators for an open source project.
The developer experience is the real draw: 700-plus integrations, deep Git sync for GitHub, GitLab, and Bitbucket, and an in-context editor that shows translators the actual UI they're translating rather than a bare spreadsheet of strings.
Pros: the lowest real entry price of any hosted tool here, genuine unlimited free tier for open source, and the deepest integration ecosystem in this comparison.
Cons: the interface leans developer-first, which some non-technical translators find less approachable than Lokalise's editor. Hosted words as a billing unit can surprise you if you add several target languages at once without checking the math first.
Who should skip it: teams that specifically want AI-scored translation quality out of the box. Crowdin has AI translation features, but Lokalise's Pro AI tier is built more explicitly around automated quality scoring if that's the deciding factor.
Why did Lokalise's pricing change so much?
This is the finding worth knowing before you commit to Lokalise: it restructured its entire pricing model between November 2025 and June 2026, and the free plan didn't survive the change.
The old lineup was Free, Start ($120/month), Essential ($230/month), Pro ($825/month), and Enterprise. The new lineup is Explorer, Growth, Advanced, and Enterprise, billed by processed words rather than stored keys or seats. Explorer starts at $144/month, Growth runs roughly $499/month, and Advanced is $999/month. There is no permanent free tier anymore, only a 14-day trial that mirrors Explorer's limits.
"Processed words" is also a meaningfully different billing unit than what Crowdin or the old Lokalise used. Every translation, re-translation triggered by a source change, and AI/MT pass counts against your quota, not just the words you store. A typo fix that triggers re-translation costs you again.
The upside genuinely exists: Pro AI on Growth and above offers automated quality scoring against Multidimensional Quality Metrics, routing content to whichever large language model fits best, and retranslation triggers when quality drops below a threshold. That's real capability. It's just capability priced for a funded team, not a solo developer testing a second language.
Pros: the most sophisticated AI translation quality tooling of anything in this comparison, unlimited translation keys on every tier, and a genuinely useful in-context editor for both web and mobile.
Cons: no free tier as of the 2026 restructure, the highest self-serve entry price here at $144/month, and processed-word billing means costs scale with activity, not just team size.
Who should skip it: solo developers and small teams on a real budget. Crowdin, Tolgee, or Weblate all cover the core job at a fraction of the cost, and the AI quality-scoring features that justify Lokalise's price only pay for themselves once translation volume is genuinely high.
What do you actually get self-hosting Tolgee for free?
Tolgee is open core, and the free self-hosted tier is a real product, not a crippled trial. Run the official Docker binaries yourself and you get core localization features, up to 10 seats, and bring-your-own-API-key machine translation through DeepL, Google Translate, or OpenAI, at no license cost.
What's held back on the free self-hosted tier: advanced features like SSO and granular permissions, professional translation ordering through partner agencies (cloud-only, since agencies need direct instance access), and any support beyond documentation and GitHub Discussions.
Tolgee Cloud exists for teams that would rather not run infrastructure. Pricing there starts around €42/month for a Team tier, moving to roughly €179/month for Business and €499/month for Advanced, with a genuinely free cloud tier at 1,000 strings and unlimited seats for small projects.
Pros: the free self-hosted tier is the real deal, not a teaser, and upgrading later from free to licensed happens by applying a key to your existing instance without losing data. Framework integrations for React, Vue, Angular, Svelte, and Next.js are solid out of the box.
Cons: 10-seat cap on the free self-hosted tier means larger open source projects will outgrow it. Support on self-hosted, even licensed, is more limited than cloud since Tolgee's team can't access your servers directly.
Who should skip it: anyone who wants zero infrastructure responsibility. If running a Docker container isn't something you want to own, Crowdin's managed free tier does more of that work for you.
Is Weblate the actual free option?
Weblate is the one tool here with no commercial tier gating core functionality behind a paywall. It's GPL-licensed, fully open source, and self-hosting gives you every feature the hosted version has, no exceptions.
Hosted Weblate exists for teams that don't want to run their own instance, priced purely on string count rather than seats or processed words, starting in the range of €19 to €299 a month depending on scale, with all plans getting the complete feature set and premium support from the core development team. Weblate has also historically offered free hosting to genuinely libre and open source projects, worth checking directly if that's your situation.
Self-hosting itself is straightforward: an official Docker Compose setup with the Weblate app, PostgreSQL, and a Redis-compatible cache container, deployable on any VPS you already run.
Pros: truly zero-cost self-hosting with no feature gate anywhere, git-native workflow that fits naturally into an existing CI pipeline, and a track record of supporting over 2,500 libre software projects.
Cons: the interface is more utilitarian than Crowdin or Lokalise, built by and for developers rather than designed for non-technical translators. Fewer polished third-party integrations than Crowdin's 700-plus.
Who should skip it: teams that need a translator-friendly interface for non-technical contributors, or that want deep integrations with design tools like Figma. Weblate's strength is the git-centric developer workflow, not the visual editing experience.
When is your framework's built-in localization actually enough?
Before paying for any of these, check whether you need to. Laravel's translation files, PHP arrays or JSON keyed by locale, handle one to three languages for free with zero external dependencies. __('messages.welcome') and a lang/ directory per locale covers a real chunk of solo products indefinitely.
The trigger to move off that setup isn't language count alone, it's coordination. The moment someone who isn't touching your codebase needs to edit a translation, or you're maintaining translation memory across five-plus languages where the same phrase needs to stay consistent, a dedicated tool starts earning its cost. Below that point, committing lang/en.json and lang/es.json to Git is the honest free answer.
The recommendation
Start with your framework's own translation files if you're under four languages. No tool, no bill, and it genuinely covers most solo products for a long time.
Once you outgrow that, default to Crowdin. The lowest real entry price among hosted tools, a free tier that isn't a trap, and the deepest integration ecosystem here.
Comfortable running one more Docker container? Tolgee or Weblate self-hosted cost nothing beyond your existing infrastructure. Pick Weblate if you want zero feature gating ever, Tolgee if you'd rather have the option to pay later for SSO without migrating platforms.
Skip Lokalise unless AI-scored translation quality is a genuine requirement. The 2026 pricing restructure moved it out of solo-developer territory, and $144/month competes with real infrastructure spend, not a nice-to-have.
If content workflows are the actual bottleneck rather than translation specifically, the Contentful alternatives post covers headless CMS options that often pair with these localization tools. And if you're evaluating your whole frontend stack while you're at it, both Astro vs Next.js vs Eleventy and Next.js vs Nuxt vs SvelteKit touch on i18n support as part of the framework comparison.
Found a better fit for your setup? Let me know on Twitter @hafizdev.
Top comments (0)