Super DeepSeek Intelegence Model since last 48 Hours

Something changed on DeepSeek side the last 48 Hours it seems. Before the DeepSeek V4 Pro was quite fast but it will go in circles for quite a time to nail a highly complex problem. Sometimes i asked me if DeepSeek is actually offloading me in the background to DeepSeek V4 flash instead serving me as DeepSeek V4 Pro. Since now 48 Hours the speed of the DeepSeek V4 Pro lowered significant compared to before but boy the Intelegence skyrocketed a lot since then for me. No matter what highly complex Problem i t