Blackwell LLM Toolkit - NVFP4 Config +Wheels + Benchmarks for Blackwell GPUs via TensorRT-LLM - 270 tk/s Nemotron 3 Omni

Was trying to get a good set of models with NVFP4 to leverage the RTX Pro 6000 and was able to get across a few hurdles and have configs + wheels set up & ran benchmarks while i was at it. hopefully this helps some folks out. This should work on all the Nvidia Blackwell cards. 5090, 5080, 5070ti etc