Product

Six modules, one machine, fully offline

Not six disconnected tools but one complete workstation: one interface, one model manager, one privacy baseline, with everything you make kept in one Library and handed from one tool to the next — and every module below keeps working just the same on a machine with no internet.

01 — Modules

Pick one to start with

Every module has its own page that takes the features apart one by one, with a demo you can play with.

MODULE 01 / 06

AI Chat

A complete AI chat experience with the model running locally — your conversations never leave your computer. Switch between open-source models, with deep thinking mode and image understanding.

Long contextContext length depends on the selected model — up to 128K. Long documents and conversations are never silently truncated.
Thinking modeTurn it on when you need deeper reasoning (off by default); the model's thought process can be expanded and inspected.
Image understandingDrag an image into the conversation or pick a file with the attach button — an on-device vision model reads it.
Document answersAttach a PDF, Word file or plain text and ask about it; the answer comes with the passages it cites.
Developer friendlyBuilt-in OpenAI- and Anthropic-compatible API endpoints, bound to localhost, ready for your own tools.

Runs100% on-device inferenceModelsMultiple open-source models, switchableContextModel-dependent, up to 128K tokens

Meet this module

AI Chat screen — Screens shown with sample data

MODULE 02 / 06

Smart Photos

Hand your photo folder to an on-device vision model: automatic descriptions and tags, one-line semantic search, people grouping and auto-curated memories — photos never get uploaded.

Two ways to searchSemantic and keyword modes, both matching the description and tags the AI wrote for each photo.
People groupingAutomatically recognizes and organizes photos of the same person, and you can name them.
Memory cardsMemory cards laid out automatically by time — browse them one by one, or save them.
Photos stay localAnalysis, indexing and search all happen on your machine.

Runs100% on-device analysis & indexingSearchSemantic and keyword modes

Meet this module

Smart Photobook screen — Screens shown with sample data

MODULE 03 / 06

Document Recognition

Advanced mode does full layout analysis with high-accuracy recognition; light mode hands whole pages to a vision model for fast text. Results feed into document comparison and insight reports.

Two modesAdvanced (layout + high accuracy) or light (fast whole-page text) — switch as needed.
Structure preservedTables, flowcharts and other structured content are not lost.
Document compareTwo documents side by side, changes at a glance.
Insight reportsSummarize and organize recognized content into reports directly.

Runs100% on-device recognitionInputScans / PDF / PowerPoint / Excel

Meet this module

Document OCR workstation screen — Screens shown with sample data

MODULE 04 / 06

Translation Studio

Not a paste-and-pray translation box: import a document, translate segment by segment against the source, refine until it reads right, then export — confidential files never touch a cloud translation service.

Segment refinementSource and translation side by side; retranslate a single segment or rewrite it by hand, and edited segments are marked.
Document importPlain text, Markdown, Word and CSV import directly, with paragraph structure preserved.
Context promptDescribe the document's field and purpose first, and the whole file is translated in that same context for consistent wording.
Fully offlineContracts, financials and research documents never leave the machine.

Runs100% on-device translationInputTXT / Markdown / Word / CSV / PDF / images / pasted text

Meet this module

Translation Studio screen — Screens shown with sample data

MODULE 05 / 06

Speech Studio

Drop in meeting recordings, interviews or videos — the on-device engine produces transcripts; videos play in theater mode with synced subtitles, and transcripts turn into summaries, mind maps and Q&A.

Audio & videoCommon audio and video formats import directly for transcription.
Synced transcriptText highlights as playback advances; click a line to jump to that moment.
Subtitle exportTranscripts export as subtitle files.
Insight reportsOne click turns a finished transcript into summaries, mind maps and Q&A.

Runs100% on-device transcriptionInputAudio / video files / live capture (Beta)

Meet this module

Speech Studio three-pane transcription workstation — Screens shown with sample data

MODULE 06 / 06

Image Generation

Generate on your own GPU: text-to-image and image-to-image with a choice of open-source models and styles, up to 2K output, everything saved to your gallery — no queues, no credits, no uploads. There is also an enhancement tool that upscales an existing image 4x and rebuilds its detail.

Text-to-image + image-to-imageStart from a sentence, or from a base image (image-to-image on supported models only).
Multiple modelsSpeed-first or quality-first — pick the model that fits.
2K outputUp to 2K: a draft is rendered first, then upscaled with its detail rebuilt — two stages combined.
Image enhancementUpscale any image 4x and rebuild its detail, then drag the divider to compare before and after.

Runs100% on-device GPU generationResolutionUp to 2K

Meet this module

Image workbench screen — Screens shown with sample data

02 — Honest Specs

The specs, stated plainly

Only a few numbers — but each one directly affects your data and your experience. Actuals depend on your device and model choices.

6Integrated modulesChat / Photos / Documents / Translation / Speech / Image

0Telemetry & uploadsNo usage statistics collected

128KContext ceilingDepends on selected model

~200 MBInstallerModels download on demand

100%On-device inferenceAI works without internet

Recommended hardware

Every module runs on standard configurations; machines with a discrete GPU and 16 GB+ RAM get the best experience for image generation and larger models. The app recommends models suited to your hardware.

Storage

Models download on demand — typical combinations take about 3–25 GB. Add or remove them anytime in the model manager, with usage shown at a glance.

Language

The interface ships in Traditional Chinese, Simplified Chinese, English and Japanese, switchable in Settings; chat and translation models are multilingual (Chinese, English, Japanese and more, model-dependent).

Updates

After 1.0, new versions download in the background and apply on next launch; model and engine updates never interrupt your work. Version 0.63.7 can also update to 1.0 right inside the app.

Got the machine? Four steps

Install, agree, activate, download models — most of the time is just the download.

Get Started