Ask questions about your documents in Private Thinking. Add TXT, Markdown, Word, PDF, or Excel (.xlsx) files to preview their contents, get a structured summary, and ask follow-up questions. Vision organizes image-based pages and complex tables into clearer content for AI analysis.

Translation now avoids adding an unsolicited spoken follow-up after the translated text, while continuing to respect your selected tone. Local model checks and status are more consistent, and switching scenarios is more immediate.

The on-device AI engine has been updated to improve compatibility for offline chat and image questions. We also refined the local AI experience, so you can keep getting useful guidance when you are without a connection.

Beiwen 1.6 makes on-device AI easier to set up and share across your apps. Download once, use across your apps The shared model center clearly shows which models are ready on this iPhone. A model downloaded by Beiwen can also be used by InspirationDraft and WristMark. Spend less time starting over Interrupted downloads can resume when possible, completed downloads are checked before use, and recovery messages make the next step clear when a retry is needed. Manage local AI with confidence Clearer installed, downloading, ready, and retry states help you know whether local AI is ready, without exposing technical setup details. More reliable downloads, less time starting over Downloads now keep your progress more reliably when interrupted, finish only when the complete file is ready, and recover safely when you return to Beiwen. Spend less time downloading again and get to on-device AI sooner.

Beiwen 1.6 makes on-device AI easier to set up and share across your apps. Download once, use across your apps The shared model center clearly shows which models are ready on this iPhone. A model downloaded by Beiwen can also be used by InspirationDraft and WristMark. Spend less time starting over Interrupted downloads can resume when possible, completed downloads are checked before use, and recovery messages make the next step clear when a retry is needed. Manage local AI with confidence Clearer installed, downloading, ready, and retry states help you know whether local AI is ready, without exposing technical setup details.

• Offline voice readout — in subways, elevators, parking garages, or any weak-signal moment, Beiwen can read AI answers aloud even without network • Ask and listen instantly — play responses with one tap so you can keep moving without staring at the screen • On-device playback path — speech readout runs locally for stable availability when connectivity drops • Interaction polish — clearer feedback for copy/readout actions and smoother handling for longer responses

• Offline voice readout — in subways, elevators, parking garages, or any weak-signal moment, Beiwen can read AI answers aloud even without network • Ask and listen instantly — play responses with one tap so you can keep moving without staring at the screen • On-device playback path — speech readout runs locally for stable availability when connectivity drops • Interaction polish — clearer feedback for copy/readout actions and smoother handling for longer responses

• Offline voice readout — in subways, elevators, parking garages, or any weak-signal moment, Beiwen can read AI answers aloud even without network • Ask and listen instantly — play responses with one tap so you can keep moving without staring at the screen • On-device playback path — speech readout runs locally for stable availability when connectivity drops • Interaction polish — clearer feedback for copy/readout actions and smoother handling for longer responses

• Transfer scenario results to general Q&A — seamlessly continue from translation, polishing, and more • Image attachments carried over when transferring context • Cleaner translation output — model-echoed instructions are now stripped

• Translation Workspace — choose a scenario and start translating right away • Image Translation — take or pick a photo; the built-in AI reads and translates text in the image • Smart composer — multi-line input with all tools at your fingertips • Language-aware translation — the system detects source language and sets the right direction automatically

• Speak your thoughts naturally — voice is transcribed and polished by on-device AI, so you get clean text without typing. • See your audio come to life with an animated waveform while you speak. • Describe attached images by voice — thumbnails stay visible while recording. • Everything stays private on your device, no internet needed.