The openHAB ChatGPT Binding allows openHAB to communicate with any OpenAI API-compatible service, e.g., OpenAI, OpenRouter, Mistral AI, Ollama, etc. and access large language models like OpenAI’s ChatGPT and many more.
This is a beta version of the official binding , implementing support for the new LLM-based HLI framework and the Chat interface in Main UI.
Changes to the official openHAB 5.2.0 binding are listed in the PR – subject to change:
main ← florian-h05:chatgpt-llmhli
opened 11:12AM - 09 Jul 26 UTC
This is a major update for the ChatGPT binding, including these changes:
- Ad… d a ChatGPTApiClient as a central facade to the API.
- Replace `apiUrl` and `modelUrl` parameters through a new `baseUrl` parameter. OpenAI API-based APIs use the same path names, swapping OpenAI against a different provider only requires adjusting the base URL - paths stay the same.
- Inherit channel configuration defaults for model, temperature, topP & maxTokens properties from the Thing's configuration.
- Fix ChatGPTModelOptionProvider not providing models for selection.
- Update the ChatGPT HLI service implementation to integrate with openHAB Core's LLM-based HLI framework and remove internal tools, context injection and history code.
- Localise error messages for the ChatGPT HLI service implementation.
- Update default temperature to 1.0 as this is OpenAI's recommendation.
Make sure to check the list of changes – things may break.
https://github.com/florian-h05/openhab-addons/releases/download/chatgpt-llmhli-0.2/org.openhab.binding.chatgpt-5.3.0-SNAPSHOT.jar
Hi Florian,
This is looking pretty great! I wonder if you could allow setting blank API key? On my LAN I do not desire LLM to require API key.
Blessings,
-Patrick
Hi,
I didn’t think of allowing a blank API key, but sure, for local Ollama it isn’t required.
I will update the binding.
I’ve just updated the binding and published a new version to the marketplace.
