Lector
lector.dev
- Free plan
- Open source
- Web
- iPhone and iPad
- API
- Self-hosted
What Lector is
Lector is an open-source language-learning reader that turns imported texts and media into reading and practice material. It accepts EPUBs, pasted text, and web articles; readers can click words for translation and track vocabulary as known, learning, or new. YouTube transcripts with timestamps and uploaded podcasts can be used for transcription and listen-along playback. Cloze exercises pair sentences by frequency and use spaced repetition. Mined words and phrases can be sent to Anki as Basic or Cloze cards, with a beta add-on that syncs reviews back. Its tutor and writing correction can use the Claude API or a local model, while AI translation can fall back to an on-device dictionary. Lector works with any language, and language packs add a frequency-ordered cloze bank and on-device dictionary. It is available on the web, supports offline access, and lists 24 languages. The Free plan costs $0 per month; podcast import and managed AI requests are excluded, and journal writing is capped at 1,000 words monthly. Paid plans start at $5/month.
Who it suits
Lector suits language learners who want to study from their own reading, web content, or media. It may also suit learners who use Anki or prefer self-hosting and local language tools.
Good to know, and what to watch
- Imports EPUBs, pasted text, and web articles.
- Tracks known, learning, and new vocabulary.
- Includes frequency-based cloze practice with spaced repetition.
- Can send mined words and phrases to Anki.
- Supports offline access and 24 languages.
- Free plan does not include podcast import.
- Free journal writing is limited to 1,000 words monthly.
- Free plan excludes managed AI requests.
Plans
Every plan and priceFree
Self-hosted
Cloud
Cloud Plus
Free plan
The verdict
Lector combines imported reading and media with vocabulary tracking and practice. The Free plan has clear limits on podcasts, journal writing, and managed AI, while paid plans start at $5/month.
What the maker publishes
- AI and language tools
- The tutor and writing correction can use the Claude API or a local model, while AI translation can fall back to an on-device dictionary. source
- AI providers
- Supported optional AI backends include Anthropic Claude and OpenAI-compatible servers such as Ollama, LM Studio and vLLM. source
- AI tutor and journal
- The built-in tutor answers grammar questions and corrects journal writing with explanations, using Claude or a local model. source
- Anki
- Lector sends mined words and phrases to Anki as Basic or Cloze cards, and its beta add-on syncs reviews back. source
- Anki integration
- Lector supports two-way Anki synchronization through its beta Lector Sync add-on, while local self-hosts can use AnkiConnect. source
- API client generation
- The API documentation gives an OpenAPI Generator command for producing a typed TypeScript fetch client. source
- API extensibility
- API settings allow provider-key storage for Anthropic or OpenRouter, model selection and OpenAI-compatible endpoint discovery on self-hosted deployments. source
- API import endpoints
- Documented endpoints import EPUB files, audio, YouTube captions and URL-extracted articles, and expose collection, lesson, progress and statistics operations. source
- Audio and video
- Users can import YouTube transcripts with timestamps or upload podcasts for transcription and listen-along playback. source
- Cloud data
- The privacy policy says cloud content is stored to provide the service and may be sent to a translation provider, and says content is not used to train models or sold. source
- Cloud region
- The homepage says Lector Cloud stores data in AWS us-east-1 and users can export it at any time. source
- Cloze practice
- Practice uses frequency-ordered Tatoeba sentence pairs with multiple-choice or typed answers and spaced repetition. source
- Company maker
- The product site identifies Luke Boyle as its builder, while his personal site says he is based in regional Australia. source
- Documentation depth
- The docs cover installation, environment variables, backups, OIDC multi-user login, Tailscale and HTTPS, reading features, language packs and the REST API. source
- Independent limit: language-pack gaps
- The Reddit post says Japanese and Korean have dictionaries and reader support but no cloze banks yet. source
- Independent limit: model dependence
- The Reddit post identifies rich translation, grammar tutoring and journal correction as the only functions requiring a model or API key. source
- Independent praise: feature parity
- The Reddit author states the cloud and self-hosted versions are the same application without an open-core split or feature separation. source
- Independent praise: installation
- A Reddit self-hosting post highlights that deployment uses one docker compose command and SQLite without an external database. source
- Independent praise: privacy
- The same Reddit post praises offline reader, cloze and vocabulary operation, with books remaining on the user's disk and no telemetry. source
- Integrations
- The documentation lists AnkiConnect, Claude or OpenAI-compatible AI servers, optional Google Cloud Text-to-Speech, and Docker. source
- Language coverage
- The site lists language packs for 24 languages including Mandarin, Hindi, Spanish, Arabic, French, German, Italian, Japanese-related Greek variants, Latin and others. source
- Language support
- The reader works with any language, while language packs add a frequency-ordered cloze bank and an on-device dictionary. source
- License
- The repository licenses Lector under GNU Affero General Public License version 3.0, including the requirement to offer corresponding source for modified network services. source
- Limit: cloud account language bug
- A GitHub issue documented a July 8, 2026 cloud bug where browser-local language state could leak across accounts and cause setup inconsistencies between browsers. source
Checked against the maker's pages on 15 Oct 2026.