Skip to main content

Autonomous Integration

Since HAP v6.1.0, it supports integration with third-party large models compatible with the OpenAI API standard. You can quickly complete model integration and invocation by configuring the base_url to the target service API address.

Configuration Steps​

Configuration steps are exactly the same as those in the based on OpenAI document. You only need to modify the following field values in the configuration file:

  • resource: The API base_url address of the target service (compatible with the OpenAI API).
  • deployment: The model name of the target service.
  • apiKey: The access credential (API key) for the target service.

Integration Examples​

DeepSeek Integration Example​

The base_url compatible with the OpenAI API for DeepSeek is: https://api.deepseek.com/v1.

Configuration file example:

md.aigc.configs=[{\
'model': 'DeepSeek-V3', \
'resource': 'https://api.deepseek.com/v1', \
'deployment': 'deepseek-chat', \
'apiVersion': '', \
'apiKey': 'sk-dd0bd***********************4306', \
'price': '0.012' \
},{\
'model': 'DeepSeek-R1', \
'resource': 'https://api.deepseek.com/v1', \
'deployment': 'deepseek-reasoner', \
'apiVersion': '', \
'apiKey': 'sk-dd0bd***********************4306', \
'price': '0.012' \
}]
  • resource: Fill in the base_url for the DeepSeek service compatible with the OpenAI API.
  • deployment: Fill in the name of the model provided by DeepSeek.
  • apiKey: Fill in the API Key provided by DeepSeek.
  • For detailed configuration steps, please refer to based on OpenAI.

Volcano Ark Platform Integration Example​

The base_url compatible with the OpenAI API for the Volcano Ark platform is: https://ark.cn-beijing.volces.com/api/v3.

Configuration file example:

md.aigc.configs=[{\
'model': 'DouBao-1.5-Pro', \
'resource': 'https://ark.cn-beijing.volces.com/api/v3', \
'deployment': 'doubao-1-5-pro-32k-250115', \
'apiVersion': '', \
'apiKey': 'da***********************5e', \
'price': '0.012' \
},{\
'model': 'DeepSeek-V3', \
'resource': 'https://ark.cn-beijing.volces.com/api/v3', \
'deployment': 'ep-2025******0241-6szgr', \
'apiVersion': '', \
'apiKey': 'da***********************5e', \
'price': '0.012' \
}]
  • resource: Fill in the base_url for the Volcano Ark service compatible with the OpenAI API.
  • deployment: Fill in the Model ID or Endpoint ID provided by the Volcano Ark platform.
  • apiKey: Fill in the API Key provided by the Volcano Ark platform.
  • For detailed configuration steps, please refer to based on OpenAI.

Ollama Local Large Model Integration Example​

Ollama provides a local service compatible with the OpenAI API, with a default base_url of: http://localhost:11434/v1.

Configuration file example:

md.aigc.configs=[{\
'model': 'Ollama-DeepSeek-V3', \
'resource': 'http://192.168.10.154:11434/v1', \
'deployment': 'deepseek-r1:32b', \
'apiVersion': '', \
'apiKey': 'example-key', \
'price': '0.012' \
}]
  • resource: Fill in the actual base_url for the local Ollama service compatible with the OpenAI API, typically in the format {host}/v1, e.g., http://192.168.10.154:11434/v1.
  • deployment: Fill in the name of the model running on the local Ollama service.
  • apiKey: Fill in the API Key for the local Ollama service. If Key authentication is not enabled, enter any non-empty string (e.g., example-key).
  • For detailed configuration steps, please refer to based on OpenAI.