DeepSeek integration

Feature Requests
21

hello , you are trying to use deepseek api keys with openai url ?

22

I didnā€™t find a way to change the url in the fork. But the original component now has it, so Iā€™m using that

23

Thatā€™s it, it should work now.
For some reason home assistant developers keep rejecting any PRs trying to add an easier option to switch the OpenAI endpoint in the official integration

  1. Install hass-environmental-variable
  2. Add this to your configuration.yaml:environment_variable: OPENAI_BASE_URL: ā€œhttps://api.deepseek.com/v1ā€
  3. Restart your system and add the OpenAI Conversation integration, when asked for the API key use the one you crated for DeepSeek
  4. Open the integration and uncheck ā€œRecommended model settingsā€
  5. Set ā€œmodelā€ to ā€œdeepseek-chatā€ and increase maximum tokens to 1024, then reload the integration
    taken from reddit
24

image
image1242Ɨ608 59.7 KB

You may also host the DeepSeek-R1 model free at your Azure AI Foundry account and connect it using the Azure OpenAI Conversation github custom component from HACS store.

Important Configuration:

  1. The base URL must include the api version (You can check this by navigating to your model + endpoint section and select your deployed model) e.g. https://ai-xxx.services.ai.azure.com/models/chat/completions?api-version=2024-05-01-preview
  1. After finish setup the service, you must click the CONFIGURE button and uncheck the recommended model settings. Click submit and configure again you will be able to change the model name to your deployed DeepSeek model (e.g. DeepSeek-R1-home) .
  2. Then, go to Settings > Voice assistants . You can then add a new assistant and select the Azure OpenAI Conversation at the Conversation Agent dropdown. This is the default name, you can change it at its integration section.
  3. Happy chatting :smiley:
25

Ideally would be a distilled version that runs locally

26

I saw this today and somehow managed to get it working by adapting code from here:

Itā€™s not perfect yet, but itā€™s working for now :slightly_smiling_face:

Just add this to your custom repositories in HACS and configure it:

Maybe a good starting point for further development!

27

Iā€™m getting api_error when entered my API key.
Do I need to top up my account to use it, or is there free tier for testing?

edit: yes, I topped up with $2 and it works now. Very slow however.

28

Since DeepSeek is open-source, many large internet companies have deployed DeepSeek services externally. Therefore, I hope the DeepSeek integration can be more open, allowing users to set different URLs and tokens themselves. In comparison, the previously mentioned OpenAI integration is somewhat lacking and appears rather closed

30

In the newest release I made the Base URL and the Chat Model configurable.
Greetings :slight_smile: