i am very disappointed in the new web/app version update

Honestly, everything was doing fine in fact, I was happy. I had seen a lot of improvements over the last days across DeepSeek’s three separate modules, especially yesterday (this week too). But somehow I wake up to find they mixed them all into one. alr, that didn’t sound bad at first it looked like a good update until I actually tried it and found out: It’s giving shallower responses, shorter ones too (even after asking for more detail it still gives lazy ones). It writes weirdly too fast (which can be both good and bad) so basically, the model’s quality dropped a lot. Its thinking is too short (literally than 3 seconds no planning like before) and full of “hmms” and “okays,” instead of the old way of thinking organized, efficient, and way better. Now it’s short and full of fluff. It’s also WORSE at following instructions now. Some people actually spent money prompting DeepSeek V4 into its current—or rather, old—way of thinking. Every new update costs prompt some engineers money. especially this most recent honely (and i am ashamed what i am saying) I didn’t ask for this update. The model didn’t get better; it got worse. It settles for “good enough” responses instead of doing all the work—and I end up doing the work by correcting it. I sent seven messages explaining issues each solved issue spawned four more, and the old ones resurfaced. I miss the old Expert already. it has bad memory now, it forgets, mixes, and literally doesn't put the effort which sounds like this model is more like 'reducing costs' instead of 'doing good work' or 'doing the hard work' Am I the only one seeing the bad change in the app/web version, or are there people who actually find it better than the old one? If so, I’m all ears.