LinguaCafe

Free, GPL-licensed language learning software — honestly reviewed. What you actually get when you stop paying for LingQ and run it yourself.

TL;DR

• What it is: Free, self-hosted language learning platform built around reading comprehension — import texts, click unknown words, build vocabulary, export to Anki [README][1].
• Who it’s for: Self-directed language learners who use the “comprehensible input” method — people who learn by reading foreign texts, not gamified drills. Especially useful for Japanese, Chinese, Korean learners who need tokenization [README].
• Cost savings: LingQ Premium runs ~$12.99–$14.99/mo. LinguaCafe is free software. A $5–10 VPS covers your full hosting cost.
• Key strength: Deep Anki integration, DeepL translation, Jellyfin media sync, and 27-language support in a single self-hosted package [README].
• Key weakness: Single-user only at the time of this review (multi-user added in v0.13 but development is ongoing), RAM-hungry if you load multiple languages, and the project is solo-developed with an active-development disclaimer warning you about bugs [README][2].

---

What is LinguaCafe

LinguaCafe is a reading-focused vocabulary acquisition tool. The core idea: you import a foreign-language text — a news article, a book chapter, a subtitle file — and read it inside a browser interface. Words you don’t know get clicked, defined, and queued for spaced-repetition review via Anki. The software tokenizes the text (splitting Chinese, Japanese, Thai into individual words, which browsers can’t do reliably on their own), tracks which words you’ve seen and marked as known, and builds a vocabulary list over time [README].

The developer describes the motivation plainly in the AlternativeTo listing: “I developed LinguaCafe mainly for personal use, because I found the alternative platforms too expensive or lacking in features I wanted” [1]. That’s LingQ. LingQ invented this reading-and-vocab-tracking approach and built a substantial business around it. LinguaCafe is the self-hosted answer for learners who want the same workflow without the monthly subscription.

It’s not a flashcard app (that’s Anki’s job). It’s not a gamified habit-builder (that’s Duolingo’s job). It’s the import-and-read layer that bridges real foreign-language content with your spaced-repetition system.

As of this review: 1,308–1,323 GitHub stars, GPL-3.0 license, 27 supported languages, built with Vue.js and Laravel, running inside Docker [README][1].

---

Why people choose it

The third-party review ecosystem for LinguaCafe is thin compared to bigger projects — no dedicated long-form reviews exist yet. What’s available tells a consistent story: learners find it via the self-hosted community and choose it specifically because LingQ’s pricing is a recurring frustration.

The AlternativeTo page shows 68 listed alternatives to LinguaCafe — and notably, people are also listing LinguaCafe as an alternative to LingQ-style tools like LinguaPal, Speeek, LinguaMerse, and half a dozen others [1]. The pattern is consistent: learners who want the “read and track unknown words” workflow are actively seeking a self-hosted option.

The selfh.st self-hosted weekly [2] covered LinguaCafe’s v0.13 release, which added support for remote MySQL servers, multi-user support, and configuration for e-book metadata import. The fact that a self-hosted news digest considered it noteworthy enough to feature signals it’s cleared the bar of being a real, usable project rather than an abandoned experiment.

Versus LingQ. This is the obvious comparison and the one the developer implicitly makes. LingQ is polished, has a massive community-created content library, and has been running for 15+ years. It also charges $12.99/mo for Premium and $25.99/mo for Premium Plus, with a free tier that caps your known words at 20 — making it useless past the beginner stage without paying. LinguaCafe has none of that. It has rough edges and no commercial content library, but your vocabulary list isn’t held hostage behind a paywall.

Versus Duolingo. Not really a comparison — different tools for different learners. Duolingo is gamified drill practice; LinguaCafe is unstructured reading. If your goal is streaks and XP, LinguaCafe won’t satisfy you. If your goal is reading a Japanese novel with manageable friction, Duolingo won’t satisfy you.

Versus browser extensions (Yomichan, Migaku). The closest free alternative for Japanese/Chinese learners is a browser extension that pops up definitions. LinguaCafe goes further: it tracks every word you’ve seen across every text you’ve imported, maintains a persistent vocabulary database, syncs with Anki, and lets you review what you’ve read in a dedicated interface. Extensions don’t persist that reading history.

---

Features

Based on the README and project documentation:

Core reading workflow:

• Text import — paste text directly or import from files [README]
• Tokenization for 27 languages, including CJK scripts, Thai, and Arabic-script languages [README]
• Click-to-define with inline dictionary lookup [README]
• Mark words as known, learning, or ignored; tracks status per word across all texts [README]
• Built-in dictionary (JMDict integration for Japanese) [README]
• Full-text search across your imported library [1]

Anki integration:

• Export vocabulary cards directly to Anki via Anki-Connect API [README]
• Cards include context sentences from the source text [README]
• This is the spaced-repetition backbone — LinguaCafe doesn’t try to replace Anki’s review algorithm

Translation and definitions:

• DeepL API integration for high-quality machine translation [README]
• Works per-word and per-sentence inside the reader interface

Jellyfin integration:

• Connect to a Jellyfin media server to import subtitles from foreign-language movies and TV shows [README]
• Lets you cross-reference vocabulary from your actual viewing, not just text imports

Vocabulary management:

• Vocabulary search and filtering [1]
• Word status tracking (new / learning / known) across your full reading history
• Statistics on vocabulary growth over time

Language support: 27 languages: Chinese, Croatian, Czech, Danish, Dutch, English, Finnish, French, German, Greek, Italian, Japanese, Korean, Latin, Macedonian, Norwegian, Polish, Portuguese, Romanian, Russian, Slovenian, Spanish, Swedish, Thai, Turkish, Ukrainian, Welsh [README]. Language support depth varies — the GitHub Wiki documents what level of tokenization and dictionary support each language gets.

Infrastructure:

• REST API [merged profile]
• Docker Compose deployment [README]
• x64 architecture; Apple Silicon works with an extra step; Raspberry Pi is not supported [README]
• RAM scales with languages loaded — can exceed 2GB if all 27 languages are loaded simultaneously (SpaCy NLP models are the culprit) [README]

---

Pricing: SaaS vs self-hosted math

LinguaCafe: Free software, GPL-3.0. You pay for hosting only.

The SaaS alternatives for context:

• LingQ Free: Caps known words at 20. Functionally unusable for serious learners.
• LingQ Premium: ~$12.99/mo. Unlimited words, full import, all languages.
• LingQ Premium Plus: ~$25.99/mo. Adds offline access and tutoring credits.
• Duolingo Super: ~$6.99–$12.99/mo depending on billing cycle.
• Readlang Pro: ~$5/mo. Good for European languages; limited CJK support.

Self-hosted on a cheap VPS:

• Hetzner CX22 (2 vCPU, 4GB RAM): ~€4.35/mo ≈ $5/mo
• DigitalOcean Basic (2GB RAM): $6/mo
• Note: the 4GB RAM tier is recommended if you’re loading more than 2-3 languages

Math for a LingQ migrant: A learner paying LingQ Premium at $12.99/mo is spending $155.88/year. Self-hosted LinguaCafe on a $6/mo VPS costs $72/year. That’s $83/year saved — not the dramatic numbers you get dumping Zapier, but it’s real money for a tool you use daily, and the savings compound if you would have eventually upgraded to Premium Plus.

The more honest framing: LingQ’s free tier blocks you at 20 words, so if you’re serious about learning a language, you’re paying $12.99/mo whether you like it or not. LinguaCafe removes that constraint entirely.

---

Deployment reality check

LinguaCafe deploys via Docker Compose. The README points to the GitHub Wiki for installation instructions. The setup is Docker-standard: clone the repo, configure the .env file, run docker-compose up.

What you need:

• A Linux x64 server with at least 4GB RAM if you plan to load multiple languages
• Docker and docker-compose
• A domain and reverse proxy (Caddy or nginx) for HTTPS if you want it accessible outside your home network
• An Anki installation with Anki-Connect plugin if you want card export
• A DeepL API key if you want machine translation (DeepL free tier gives 500K characters/mo)
• A Jellyfin server if you want subtitle import from your media library

What can go sideways:

The README includes an explicit caveat: “LinguaCafe is still in active development, you might encounter some bugs while using the software. Please test it before you start actively using it.” [README] This is honest and worth taking seriously — it’s a solo-developer project, not a company-backed product.

The ARM limitation is a real constraint: Raspberry Pi 3 and newer are explicitly not supported [README]. If your homelab is ARM-heavy, you’ll need an x64 machine or a VPS.

Memory usage is the other practical concern. SpaCy language models are loaded in RAM — if you’re learning Japanese and Korean simultaneously, plan for 2–4GB RAM just for the NLP layer. The README states memory “can be over 2GB” if all languages are loaded [README].

Multi-user support was added in v0.13 [2], but the README still notes “only one user/server is supported” — suggesting the feature was added but may still have rough edges at the time of writing. If you’re planning to share an instance with a study partner or family members, test this before committing.

Realistic setup time: 30–60 minutes for someone comfortable with Docker on Linux. For a non-technical learner following a guide: 2–3 hours including domain setup. The main blocker is usually Anki-Connect configuration, which requires installing the Anki desktop app and an add-on before card export works.

---

Pros and Cons

Pros

• Genuinely free. No word limits, no subscription tiers, no features locked behind payment. GPL-3.0 means you can audit the code and run it indefinitely [README][1].
• Anki integration. Vocabulary goes straight into the world’s best spaced-repetition system with context sentences. You’re not stuck with a proprietary review mechanism [README].
• DeepL quality. Machine translation at LinguaCafe’s level is better than the in-page Google Translate most alternatives rely on [README].
• Jellyfin sync. If you’re already self-hosting your media and watching foreign-language TV, importing subtitles directly into your vocabulary tracker is a genuinely useful workflow that no paid competitor offers [README].
• 27 languages. Covers most major European and Asian languages including less-served ones like Welsh, Macedonian, Slovenian, and Croatian [README].
• Data sovereignty. Your reading history, vocabulary lists, and progress data live on your server. LingQ has had pricing changes and feature deprecations over the years; your LinguaCafe data doesn’t go anywhere.

Cons

• Solo-developer project. One maintainer (simjanos-dev), no company backing. If they stop developing it, the project stalls. No SLA, no support contract, no guarantee of continued development [README].
• Active development = bugs. The developer’s own readme warns you to test before committing. For a daily learning tool you’ll use for years, that’s a meaningful caveat [README].
• No content library. LingQ’s biggest practical advantage is 20+ years of community-uploaded graded content and mini-stories. LinguaCafe has no built-in content. You source your own texts [README].
• No mobile app. LingQ has iOS and Android apps. LinguaCafe is browser-based. Reading on your phone requires accessing the web interface, which works but isn’t native.
• ARM not supported. Raspberry Pi 3 and newer explicitly won’t work. x64 VPS required [README].
• RAM appetite. 2GB+ if loading multiple languages — more than many basic self-hosted tools require [README].
• Multi-user is new and unverified. Added in v0.13 [2], but the README disclaimer hasn’t been updated. Shared instances may have rough edges.
• No community content. Unlike LingQ, there’s no sentence library or shared vocabulary lists to import. You’re building your content pipeline from scratch.

---

Who should use this / who shouldn’t

Use LinguaCafe if:

• You’re paying for LingQ and the bill bothers you.
• You already have a Docker-capable machine (NAS, homelab, cheap VPS) and are comfortable with basic server administration.
• You’re serious about comprehensible input — you read in your target language regularly and want to systematically track and review vocabulary.
• You use Anki for spaced repetition and want your reading vocabulary to flow automatically into your deck.
• You’re learning Japanese, Chinese, Korean, or Thai where proper tokenization matters for any vocabulary tracking tool to work.
• You have Jellyfin and watch foreign-language content — the subtitle import workflow is uniquely useful here.

Skip it (consider Readlang or a browser extension) if:

• You want a minimal setup with zero server overhead. Readlang is $5/mo and needs nothing installed.
• You’re learning a single European language and Yomichan-style browser extensions cover your use case.
• You’re a beginner — LinguaCafe’s reading-first workflow assumes you can already read some of the target language. It’s not a structured curriculum.

Skip it (stay on LingQ) if:

• You rely on LingQ’s graded content library and mini-stories — no self-hosted equivalent exists for that.
• You need a polished mobile app for commute study.
• You’re not comfortable managing a VPS and have no one to help.
• The $12.99/mo doesn’t bother you and you’d rather someone else handle maintenance.

Skip it (use Anki alone) if:

• You want spaced-repetition only and you source your cards from pre-made decks. LinguaCafe’s value is in the reading-and-capture workflow; if you’re not importing original texts, it adds overhead without adding much.

---

Alternatives worth considering

• LingQ — the original. Polished, mobile apps, 20+ years of community content, active streaming import. $12.99–$25.99/mo. The reference point LinguaCafe was built against.
• Readlang — browser extension plus web reader, $5/mo. Good for European languages; CJK tokenization is weaker. Easier setup than self-hosting.
• Yomichan / Yomitan (Japanese) — browser extension for Japanese with excellent dictionary integration and Anki export. Free, no server needed, Japanese-only. If you only need Japanese and don’t care about persistent reading history, this covers 80% of what LinguaCafe does.
• Migaku — browser extension with SRS integration, targeting the same audience. Subscription-based. Better media integration than Yomichan but less flexible.
• Anki alone — if you already have discipline around creating cards from your reading, you may not need the reading-layer software at all.
• LibreLingo — open-source Duolingo clone, completely different approach (structured lessons vs. reading). Not a real alternative for this use case.

For the self-hosted learner community, the realistic choice is LinguaCafe vs. just paying for LingQ. If you’re comfortable with Docker and already have a server, LinguaCafe is the obvious move. If you’re not, LingQ is the path of least friction.

---

Bottom line

LinguaCafe fills a specific gap cleanly: it’s the self-hosted, free-forever version of the workflow LingQ commercialized. If you’re a serious language learner who reads in your target language regularly and you’re tired of paying $12.99/mo for the privilege of tracking vocabulary across your own imported texts, LinguaCafe is a credible replacement. The Anki and DeepL integrations are genuinely useful, the 27-language support is broad, and the Jellyfin sync is a differentiator no paid tool matches.

The trade-offs are real: it’s a solo-developer project with active-development warnings, no content library, no mobile app, and RAM requirements that rule out lower-end ARM hardware. But for the learner who already self-hosts and already uses Anki, the setup cost is low and the ongoing cost is a $5 VPS. That’s the math that keeps showing up in self-hosted learning communities, and it’s why this project has accumulated 1,300 GitHub stars despite no marketing budget and no company behind it.

If the setup is the blocker, upready.dev deploys self-hosted tools like this for clients. One-time fee, you own the instance, no recurring bill.

---

Sources

1. AlternativeTo — LinguaCafe (5 likes, 68 alternatives listed, GPL-3.0, 1,323 GitHub stars). https://alternativeto.net/software/linguacafe/about/
2. Ethan Sholly, selfh.st — “This Week in Self-Hosted (28 June 2024)” — covers LinguaCafe v0.13 release with multi-user support and remote MySQL. https://selfh.st/weekly/2024-06-28/

Primary sources:

• GitHub repository and README: https://github.com/simjanos-dev/linguacafe (1,308 stars, GPL-3.0, solo maintainer)
• Official website: https://simjanos-dev.github.io/LinguaCafeHome/
• GitHub Wiki (user manual and setup guide): https://github.com/simjanos-dev/LinguaCafe/wiki