Hey folks! With Claude 4 Sonnet just released today, I'm curious about how it stacks up against Claude 3.5 Sonnet, especially for coding tasks. Has anyone had a chance to try both? What are your thoughts?
4 Answers
Honestly, it’s not even a competition—Claude 4 Sonnet completely outshines Claude 3.5. The improvements are undeniable!
I've been using Claude 4 for coding after typically going with 3.7 and 3.5 for different tasks. I can definitely say it’s a step up for coding, but be cautious if you’re on a limited plan! It tends to generate longer responses by default, which could burn through your quota.
To be fair, I actually prefer 3.7 over 4. However, in specific cases like Claude Code, 4 does perform better than 3.7. It can be a mixed bag depending on what exactly you're working on.
I've been kind of let down by Claude 4 Sonnet. Everyone seems to love it, but I'm not having a great experience. I used Claude 3.5 for coding, and it really respected my coding style and constraints. Now, it feels like 4 just ignores those rules, which makes it hard to incorporate into my workflow. I plowed through my quota trying to make it work and now I’m here just waiting for my quota to reset.
I totally get that frustration! Claude 3.5 was brilliant for coding because it could adapt to your existing styles. I wish they'd bring back that functionality in 4, but it seems they’re only focusing on the benchmarks now.