4 ms·
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
by themoonisachees 3y ago
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 comparable
- kingsleyopara 3y agoAt least in my tests, GPT-4 also seems suspiciously good at reading text so it’d be surprised if they weren’t also doing this.
- Departed7405 3y agoGPT4 can't read non-latin languages (OpenAI admits it) such as Persian, or Chinese. It will halucinate meaning. Gemini Ultra nails it.