Claude Sonnet 3.7 and ChatGPT 4.5 are very close in performance now. I still maintain that GPT is a bit more 'structured and analytical' (formal, if you will) while Claude retains a feeling of creativity, perhaps more of a personality. - Which is why I prefer to ideate with it. To get the best from both - I simply pit them against each other when reviewing and refining work. Simple steps: 1. Start in Claude (it's a bit more 'creative') - it works better with broader ideas and keeps possibilities open before narrowing down 2. Give the same task to GPT 3. (likely) the GPT version is a bit too concise or restrictive - ask Claude to evaluate and highlight to you the bits it would use or are improvements 4. Review and get Claude to update it's thinking based on your direction 5. Share this to GPT and ask it to do better based on what Claude did 6. Take this 'best from GPT' back to Claude to get a final version* or confirmation you've reached 'peak' *repeat 6 and 7 until you are satisfied with the output*