Open-model launches used to create a short, familiar cycle. First came the benchmark discourse, then the GitHub issue pile, then the inevitable week or two where everyone pretended deployment details were somebody else’s problem. NVIDIA is trying to kill that gap. Its same-day DeepSeek V4 post is not really