Deepseek V4.1 is just mindblowing

I just finished some generations to aggregate technical context into human readable paper output. All previous attempts with qwen3.8 max, qwen3.8 flash next, GLM 5.3 flash and DS4 flash was struggling yesterday. I just tried again today with the same skill, and i was just amazed about the result. Everything was clear, well explained, easy to follow and the model produced something beyond my expectations. It's amazing speed turned the generation time from like 30min+ for 3.8 max to less than 5 min, and the price is just ridiculous. Intelligence distribution is a very important factor to see the impact of a model once released. And i think this one will change a lot of things. Very excited to see how people will use this cheap fast and high quality intelligence in the near future. Also interested to know what do you guys think of it in your workflows ?