v0.12.67+12067-macos

BasedHardware/omiv0.12.67+12067-macosJul 10, 2026by mdmohsin7

AI Summary

A major stability update fixing AI response timeouts, conversation caching, health diagnostics, brain map performance, and voice transcription reliability.

Key Highlights

  • Fixed stalled AI chat responses with timeout messages
  • Improved conversation caching to load instantly without hiding details
  • Added structured health-event snapshots for diagnostics
  • Improved brain map reopen speed
  • Enhanced voice and transcription reliability

New Features

  • AI chat timeout handling
  • Conversation caching isolation
  • Health diagnostics export
  • Brain map performance
  • Update safety with explicit channels
  • Voice reliability improvements

Full Release Notes

## OMI Desktop v0.12.67

### What's New
- Fixed a stalled AI chat response silently ending with no message — it now shows "Response took too long. Try again." so you know to retry
- Improved Conversations so cached details load instantly without hiding newer summaries, statuses, titles, stars, or folders
- Kept conversation details and edits isolated across account changes while preserving simultaneous title, star, and folder updates
- Desktop flaky paths now fail soft with structured health-event snapshots for diagnostics export (auth, WAL, agent bridge, API retry, DB lock backoff, mode-switch waits, BLE decode, automation bind)
- Fixed the This device memory filter for new desktop conversations and memories
- Made the desktop brain map reopen faster by preserving its scene during navigation and reusing cached layouts when the graph has not changed.
- Fixed Reset Onboarding clearing signed-in state incorrectly and leaving the app in a broken half-onboarded state
- Preserved unsent chat drafts separately across the main app, notch, agents, tasks, restarts, and updates
- Made desktop updates safer with qualified beta releases, explicit stable and beta channels, and verified post-relaunch install reporting
- Made push-to-talk and realtime voice more reliable during barge-in, provider delays, playback fallbacks, and rapid repeated turns.
- Improved transcription cloud/local fallback reliability and memory sync stability.

### Downloads
- **DMG Installer**: For fresh installs, download the DMG below
- **Auto-Update**: Existing users will receive this update automatically via Sparkle

<!-- KEY_VALUE_START
isLive: false
channel: candidate
edSignature: ZVrrBr+PD7bLvoJzWzPZkZt+DJMDLaAzEcUAHRPRXAcyKJ2gG6++nNXHeNUqMJrjEdIx0XsR2yxRKE0BrjYrAA==
changelog: Fixed a stalled AI chat response silently ending with no message — it now shows "Response took too long. Try again." so you know to retry|Improved Conversations so cached details load instantly without hiding newer summaries, statuses, titles, stars, or folders|Kept conversation details and edits isolated across account changes while preserving simultaneous title, star, and folder updates|Desktop flaky paths now fail soft with structured health-event snapshots for diagnostics export (auth, WAL, agent bridge, API retry, DB lock backoff, mode-switch waits, BLE decode, automation bind)|Fixed the This device memory filter for new desktop conversations and memories|Made the desktop brain map reopen faster by preserving its scene during navigation and reusing cached layouts when the graph has not changed.|Fixed Reset Onboarding clearing signed-in state incorrectly and leaving the app in a broken half-onboarded state|Preserved unsent chat drafts separately across the main app, notch, agents, tasks, restarts, and updates|Made desktop updates safer with qualified beta releases, explicit stable and beta channels, and verified post-relaunch install reporting|Made push-to-talk and realtime voice more reliable during barge-in, provider delays, playback fallbacks, and rapid repeated turns.|Improved transcription cloud/local fallback reliability and memory sync stability.
KEY_VALUE_END -->