Right now, we have only plain run times, model sizes etc. Ths goal is to have benchmarks that compares different tools like ONNX, MLX, etc.