VisionBridge proxy gives text-only LLMs vision capabilities

VisionBridge is an MIT-licensed OpenAI-compatible proxy that adds vision to text-only LLMs via a separate vision model. It supports tools like look, OCR, scan, crop, and compare without training.