Tap the image upload button below to attach photos. Image prompts automatically use the best working Vision model, with automatic fallback if one model is busy.