v1.1.0
Sharper control over the things you run yourself — your local model and your Telegram users — plus a round of reliability fixes for installs, voice notes, and reconnecting assistants.
Highlights
Control your Telegram users
Decide exactly who can message your assistant on Telegram with a new allowlist you manage right from the app.
Eyes on your local model
A new Inference Server panel shows your local model's status and lets you restart it in one click if it ever crashes.
- Feature Show local model inference server status in the Settings page.
- Feature Add support for Telegram allowlist mode.
- Fix Fixed fresh local-model installs that could leave a new assistant unable to start.
- Fix Keep VM connection stable during network hiccups.
- Fix Ensure voice transcription continues working after the macOS device goes into sleep mode.
- Fix Fixed a "Couldn't connect" screen that could appear after replacing an assistant.
- Other Properly display transition states (Starting / Stopping).
- Other Transcription model selector is now simpler and shows download progress in Settings.