However, the article also points out some issues with Gemini Ultra. It sometimes struggles with large pasted code and does not yet support file upload. There are also inconsistencies in its responses depending on how a question is phrased, leading to confusion about its capabilities. Some users have also raised concerns about potential bias in Gemini's image generation, although others have not experienced this issue. Despite these issues, the general consensus is that Gemini Ultra is a useful tool, especially for personal projects.
Key takeaways:
- Users find Gemini Ultra comparable to GPT-4, with some appreciating the 'check sources' feature, though it only works about 25% of the time.
- For obscure queries, Gemini is often better than GPT-4, but GPT-4 is preferred for ML Python coding as Gemini sometimes adds unnecessary elements to its code.
- Some users subscribe to both services and choose the answer they prefer for tricky programming questions. However, Gemini's inability to handle large pasted code and lack of file upload is a drawback for some.
- Gemini Ultra has unique features like inline image responses and a different drawing style compared to DALL-E, but some users still prefer ChatGPT for general text responses.