This also matches my experiences using the major LLMs. For professional work, Claude Sonnet 3.7 is clearly superior, and not just for code but pretty much any knowledge or technology related task.

I’ve been recently comparing a lot of different models on coding tasks for work. It’s the worst work I’ve ever enduring in my life. It’s like trying to convince the most eager and literate goldfish to program. They’re occasionally very helpful, but for the most part they’re just more work than they’re worth. Maybe they can save me effort in the long run, but it would seem only at the cost of my patience and sense of satisfaction with my work.

I will say my own experience had me ranking Sonnet 3.7 below Gemini 2.5 tho

TL;DR Claude 3.7 Sonnet is the best in terms of pure quality and aesthetics. O1-Pro is the worse (both in terms of quality and cost). DeepSeek V3 and Google's 2.5 Pro are the most budget-friendly options. Grok is grokking.