v0.12.286+12286-macos

Michael-A-Kuykendall/shimmyv0.12.286+12286-macosSep 4, 2026by mdmohsin7

AI Summary

This release focuses on refining voice interactions and permission handling. It fixes issues with permission prompts, prevents duplicate transcript entries, and ensures voice replies are processed cleanly.

Key Highlights

Voice transcripts no longer duplicate Omi's spoken replies.
Permissions denied during onboarding are respected and not re-asked.
Omi no longer reads Interject's classification aloud.
Push-to-talk replies are not re-corrected if fully heard.

Full Release Notes

## OMI Desktop v0.12.286

### What's New
- Omi no longer reads Interject's classification aloud during voice replies
- Skipping or denying a permission during onboarding no longer makes Omi re-ask for it on every launch, app switch, or settings sync — microphone, screen recording, and file indexing stay off until you turn them on yourself
- Push-to-talk replies you fully heard are no longer repeated or re-corrected on later, unrelated questions
- Voice transcripts no longer duplicate Omi's own spoken replies picked up by the microphone

### 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: osfaWqQvZa8tS5ZdwS2jpwAJtDhqe2O/ov/EtIvfF6J2QwvxexoNooHfA8OSXn7bDcJ9QPd08QpGlJBj7f04CA==
betaEdSignature: bDdY8tWGDQj6WGdGZYBLkNkme8Fv15uuFIishk+ZCf1eP1FxrimgCkFGFexzaMMs1tVediSZmriXPofIpQP4Bw==
changelog: Omi no longer reads Interject's classification aloud during voice replies|Skipping or denying a permission during onboarding no longer makes Omi re-ask for it on every launch, app switch, or settings sync — microphone, screen recording, and file indexing stay off until you turn them on yourself|Push-to-talk replies you fully heard are no longer repeated or re-corrected on later, unrelated questions|Voice transcripts no longer duplicate Omi's own spoken replies picked up by the microphone
KEY_VALUE_END -->