v0.3.8 Support for Frigate

Changelog

Features

• Added support for image entities like the ones provided by frigate.
  You can now analyze image entities directly! No need to save them as files and submit those anymore
• The new image entity input can also be mixed with the old image_file input
  so you can submit multiple images from different inputs at once

Bug fixes

• Fixed some error messages not appearing correctly so you have an easier time debugging

v0.3.7

Changelog

• #9 Added the detail parameter for OpenAI and lowered the minimum for image_width to 512px. This gives you more control over how many tokens a request will use.
• For now this only applies to OpenAI and will not have an effect on other providers.
• Better error logging

v0.3.6 Small Improvements

Changelog

• #9 Added the detail parameter for OpenAI. This gives you more control over how many tokens a request will use.
  For now this only applies to OpenAI and will not have an effect on other providers.
• Better error logging

v0.3.5 Support for Ollama

Changelog

This release adds another provider: Ollama

• #4 Support for Ollama
• Various small optimizations

v0.3.0 Support for LocalAI

⚠️ You will need to set up your integration again after this update!

This release is a complete rewrite of the integration adding flexibility to add more AI providers in the future. For now OpenAI and LocalAI are supported. You will have to set up providers separately.
Additionally, this integration now supports mulitple images per call for more context.

Changelog

• Added LocalAI as provider
• Support for multiple images in the same call
• temperature parameter has beed added
• More helpful error messages and input validations have been added

If you encounter any bugs, feel free to open a bug report.

v0.2.2 Performance Improvements

Changelog

• Switched to homeassistant's built-in requests aiohttp_client, which handles requests asynchronously.

v0.2.1 - New UI Setup Process

⚠️ Breaking Change: Please read before updating

For existing users:

This update introduces a cofig flow. Which means the integration can now be set up entirely from the UI, no need to add you api key to your configuration.yaml.
You'll need to set up this integration again through this new UI setup process.

After this update:

1. Remove the gpt4vision: and it's api: in your configuration.yaml file
2. Restart Home Assistant
3. Set up the integration in your Home Assistant settings:
4. Add Integration
5. Search for "GPT-4 vision"
6. Paste your API key
7. Submit

For new users:

• Enjoy the new and improved setup process.

Changelog

• #1 Setup is now done in the UI
• Various code optimizations
• Support for translations (coming soon)

As always, if you have any problems, feel free to open an issue or PR

v0.1.4 Better error messages

Changelog

• Added error handling so it is now easier to find out what went wrong
• Added an icon for the service