Deepseek-r1 thinks for 30 minutes?

I was trying to ask a question about coding using DeepSeek-R1-0528-Qwen3-8B-Q4_K_M, and the thinking took 30 minutes???

https://preview.redd.it/kex3fgg4lgvg1.png?width=277&format=png&auto=webp&s=5f7e7cdc8502b935ea8b8fb83e0e4af60c3c4533

I had to manually stop if because it just kept going.
Is there any way to mitigate this to generate only like ~2 minute thinking?
using lm studio