Has anyone tested and compared the performance of the new version with Sonnet and Deepseek-R 1?

[crazygo]

I see that the existing tests were based on GPT4o and were conducted 8 months ago.

Now that we have better code models like Claude 3.5 Sonnet and Deepseek 3.5, I'd like to know how these two models perform on different frameworks, especially on Aider.

Thank you.