- Added Traditional Chinese: the interface and the AI's steps and replies now support 繁體中文 - Fixed Memory notes written in Traditional Chinese being rewritten in simplified characters

- Added DeepSeek, refreshed the online model list to the current Claude and GPT models, and you can now enter a model ID yourself - Fixed getting stuck on the model setup screen when the online model you picked was no longer in the list, and your lists are readable from that screen now - You can mark a list as done from the home screen without ticking every step yourself, and in the Done list you can swipe to delete or start a list again from the long press menu - Deleting a list and marking one as done now animate - Fixed the answer language for very short input, for French and Spanish, and when your saved Memory notes are in a different language from the task you typed - Tightened the spacing above the steps inside a list

- Fixed an issue where Memory notes sometimes didn't take effect - Fixed the app language setting not always taking effect, mainly affecting English users - Fixed the cursor jumping while typing a new task - Fixed the keyboard closing when tapping the conversation header - Improvement: steps now answer in the language you type: English in, English out

We removed the accidentally added sound effects — it is haptic feedback only now.

Got an idea, a bug, or a wish? Now you can send it straight from the app. • Feedback & Requests, new in Settings. Write in any language, we read everything. • Plus a quick way to rate Small Todo on the App Store. And a steadier, more responsive chat under your steps.

Now you can add steps yourself and copy any message. • Add a step by hand — just type it into the chat. • Long-press a message to copy it. Plus a smoother, steadier chat as you type and rearrange your steps.

Minor improvements to the chat: long conversations can now fold away to keep your steps in view.

Clearer model choices, and a way in when your iPhone can't run one. - Bring your own model: add an API key in Settings and Small Todo works even without Apple Intelligence or an on-device model. - Your key takes effect the moment you save it, so you go straight to your lists. - Settings names the model this device can run, and explains the ones it can't. - Clearer messages when an online model is busy or the model you picked is unavailable.

Now your breakdowns keep going, even when you step away. - Background breakdowns: leave, switch apps or quit, and they carry on, then resume where they left off. - The new-list box grows as you type, with a fresh prompt each time. - Smoother typing across languages, Chinese included. - Smarter models: a nudge toward the right one, one-tap switching, and a check before you remove a download. - Steadier downloads that verify as they go and push through a shaky connection.

- More devices will be able to use Qwen models now - Add online model support for advanced users (you need to provide API Key - "bring your own key"). - Lots of experience improvements PS: When using online models, we still cannot read any of your data, AI service providers (OpenAI / Anthropic) will see your requests.

What's new: a smarter model and steadier downloads. - New Qwen3.5 (recommended), smarter and more capable. Pick from three models (Apple Intelligence, Qwen3.5, Qwen3) and switch anytime in Settings. - Sturdier downloads: they pick themselves back up after a dropped connection, pause and resume on demand, and show progress right at the top. - Finished lists tuck themselves away, so Home stays focused on what's still left. - Plenty of small polish: a snappier keyboard and a smoother thinking animation.