ParentFull threadthemoonisachees·Pretty sure gpt-4 is doing image-to-llm directly while Gemini also runs the image through Google lens which is tailor-made for OCR. I think that's a good thing to create a better product overall but the performance isn't really comparableView on HN