willitrun CLI checks model fit, speed

willitrun solves the "download and pray" problem for local AI, moving from theoretical TFLOPS to empirical benchmark-driven predictions.

• –Combines real-world benchmark lookups with scaled fallback estimations for 482+ benchmarks across 88 devices
• –Prevents wasted time and bandwidth by flagging models that will exceed VRAM or run at unusable speeds
• –Extends beyond LLMs to cover vision models like YOLO and ResNet on edge hardware like Jetson Orin
• –Supports offline use and HuggingFace integration, making it a versatile tool for both cloud and local developers
• –Community-driven benchmark updates ensure data stays fresh as new models and hardware configurations emerge