Chose between GPT-4 and Claude for a side project last month

I was building a little tool to help me summarize customer reviews for my eBay store and had to pick between GPT-4 and Claude 3.5 Sonnet. Ended up going with Claude because it handled longer text blocks without cutting off and the output felt more natural lol. Spent about 3 hours testing both with the same 50 reviews and Claude got the tone right way more often. Anyone else find one works better for specific tasks or is it just me?

2 comments

2 Comments

patriciam2217d ago

You said "the bigger picture right" about Claude. Honestly though, isn't a summary tool just that - a summary? Not sure why it needs to "get" the bigger picture like it's some deep thinker. Both models spit out text based on training data. One isn't smarter than the other just because it sounds smoother to you. Maybe the real difference is just how they handle formatting and token limits. People act like picking between these things is some huge philosophical choice when it's really just which one bugs out less on your specific use case.

iris_rivera4422d ago

That thing about handling longer text blocks without cutting off is exactly why I picked Claude for summarizing property lease agreements. The tone and flow just felt more natural to me too, especially when I was trying to get a clear plain English version of some dense legal stuff. Honestly, GPT-4 seemed fine for shorter stuff but Claude just got the bigger picture right.