Enables LLMs like DeepSeek to understand images by calling external vision models via OpenAI-compatible API. Provides tools to describe images or diagnose connectivity.