You could run it in a loop, asking it to improve the code each time. I know what the ffmpeg devs have done is impressive, but I would be curious to know if something like Claude 4 Opus could make any improvements.
I think if it was easy for them to improve critical projects like ffmpeg, we'd have seen some patches that mattered already. The only activity I've seen is LLMs being used to farm sec-ops bounties which get rejected because of poor quality.
https://daniel.haxx.se/blog/2024/01/02/the-i-in-llm-stands-f...