v1.7.0 Beta 1 Release Notes
We’ve released our own AI model, Glimpse-v1! While general multi-modal LLMs can describe images well, they often miss the contextual intent behind a scene. Sending every motion event to a cloud LLM is slow, costly, and exposes private footage to third parties.
We solved this by post-training Gemma 3 with reinforcement learning on thousands of reward-labeled samples. Through this process, our model has learned that “a person wearing a blue uniform carrying a package, with a gray van parked near the driveway”, is actually an “Amazon delivery”. Glimpse also understands the difference between a false positive (triggered by motion), and an actual event.
Additionally, the model was trained to generate event title and description in a single call. This significantly reduces inference time, so notifications can be sent more quickly.
We believe privacy is fundamental to smart homes. As such, we encourage everyone to run the AI models needed for LLM Vision on a local machine. By building specialized, compact models that can run locally on hardware with limited memory and compute resources, we aim to make local AI accessible to everyone.
While we’re excited what’s possible in the future, this is the first release of our model. As such, there will be limitations. For example, the model can currently only generate English responses.
We need your help! Models like these need a lot of training data to understand different situations well. You can help make Glimpse better for everyone by submitting feedback through the LLM Vision Card.
Check out Glimpse-v1 on our website: https://llmvision.org/glimpse/
Feedback is entirely optional, and never happens without your explicit permission. We do not collect any personal data. See our privacy policy.
Contributors
A huge thank you to our contributors
Integration
✨ Features
- Support for Glimpse-v1: This release adds optimizations for our own model which generates title and description in a single call. (by [@valentinfrlch](https://github.com/valentinfrlch))
- New action: In 1.6.0 we introduced the new timeline API. Now, in addition to the API, you can fetch events from your timeline based on filters via a new Home Assistant action: get_events. (by [@valentinfrlch](https://github.com/valentinfrlch))
- Timeline API: Now supports optional
startandendparameters to query the timeline with a precise range. The limit has also been increase to 10’000. (by @Wysie) - Configurable reasoning effort: Some providers now support configurable thinking/reasoning effort. (by [@valentinfrlch](https://github.com/valentinfrlch)) ([#609](#609))
🔧 Fixes
- Bring back calendar attributes: The calendar extra attributes have been deprecated in favor of the new timeline API in 1.6.0. For ease of use we're bringing back some calendar extra attributes (for the most recent event). To fetch events based on filters, see the new get_events action. (by [@valentinfrlch](https://github.com/valentinfrlch))
- Timeline refresh: The timeline now refreshes correctly. (by @CamSoper), (#600, #613)
- Default Google model: The default model for the Google provider has been updated to gemini-3.1-flash-lite-preview (by [@valentinfrlch](https://github.com/valentinfrlch)) ([#622](#622))
- Test coverage: We've drastically increased test coverage from <10% to >80% in an effort to make llm vision more stable and dependable. (by [@valentinfrlch](https://github.com/valentinfrlch))
🌐 Languages
- Support for Danish: Added 🇩🇰 Danish translations. (by [@Dinnsen](https://github.com/Dinnsen))
- Support for Greek: Added 🇬🇷 Greek translations. (by [@teofanis](https://github.com/teofanis))