DeepSeek DeepThink Taking Too Long? Is It Stuck or Normal?

Bit of both. The thinking mode really does take longer, that's the whole idea, it works through the problem before answering instead of blurting out the first thing. For a hard problem, a couple of minutes is normal.

But if it's going past that and then just stops with nothing, that's the server dropping you, not the model still working. It happens a lot at busy times.

My rule now is DeepThink only for stuff that genuinely needs it, tricky logic, a maths problem, an algorithm I can't figure out. For "write me a function that reads a CSV" the normal mode is faster and just as good. No point waiting 3 minutes for something a normal reply handles in 10 seconds.