OrbitQuant is a data-agnostic weight-activation quantizer designed to reduce the high inference costs of Diffusion Transformers (DiTs). Unlike traditional post-training quantization (PTQ) methods that require re-fitting calibration data due to shifting activations, OrbitQuant bypasses range estimation to handle varying timesteps, prompts, and modalities. This approach optimizes efficiency for state-of-the-art image and video generation models.