Where is AWQ used in practice?

AWQ is particularly popular for serving 70B-parameter models on consumer GPUs and for compressing models to fit on edge devices. AI governance teams adopting AWQ document the quantization configuration alongside the base model because 4-bit AWQ produces meaningfully different outputs from FP16 baselines on some inputs. The platform stren