Configure the Model

AutoLXB needs an LLM / VLM endpoint compatible with the OpenAI Chat Completions style API. Visual execution and semantic route adaptation require image understanding.

Entry

Open: Config -> Device-side LLM Config

Fill in:

• API Base URL
• API Key
• Model

The config page shows the final resolved request URL in real time, which helps you confirm whether the /chat/completions endpoint is correct.

Test the model

After filling in the fields, tap the test button.

A successful test means:

• The app can reach the model service.
• The API key and model name are usable.
• The model can process image input.

Choose a model with image understanding

If the model only supports text, a basic connection test may pass, but visual execution, screen observation, and semantic adaptation will not work properly.