Analyze any image URL using GPT-4o-mini vision. Returns structured analysis based on the mode: describe (full description), ocr (text extraction), chart (data/trend extraction), ui (interface analysis), identify (object/subject ID), or qa (answer a specific question about the image). Input must be a publicly accessible image URL (JPEG, PNG, GIF, WebP). $0.050/call.