🆕 Think button: A new "Think" button will automatically switch to a (reasoning) variant of your current model if applicable.
🆕 Lazy-loaded thread list: Support for the new Kagi Assistant paginated thread list and search API. Faster load times with infinite scrolling (better than the web app 🙃).
🆕 Background generation: Assistant keeps working and pings you when your reply is ready when you've switched apps.
🆕 Interrupt on demand: The send button becomes a stop button during generation to pre-maturely interrupt the model.
🆕 Retention setting adherence: Automatically syncs your cloud thread retention preference (24h or forever) and follows it for new threads.
🔧 Tons of perfectionist tweaks to UI/UX, as always.
What's Changed
- Fix model picker pushing send button off screen by @m2jest1c in #50
- Support for new kagi pagination API by @httpjamesm in #52
- feat: fetch and account for AUTO_SAVE by @httpjamesm in #54
- Stop active generation by @httpjamesm in #53
- Foreground service streaming by @httpjamesm in #55
- feat: auto internet setting by @httpjamesm in #56
- style: sup tweaks for better UI by @httpjamesm in #57
- feat: separate about screen by @httpjamesm in #58
- feat: disable metadata btn when none to display by @httpjamesm in #60
- fix: less jarring UX during search by @httpjamesm in #61
- feat: auto dismiss keyboard when thread drawer opens by @httpjamesm in #62
- fix: eagerly request notif permission sync by @httpjamesm in #63
- Think button UI/UX abstraction by @httpjamesm in #64
- style: fully round user messages by @httpjamesm in #65
New Contributors
Full Changelog: v1.9...v1.10