getResolvedGpuLayers abstract method
Returns resolved GPU layers used for the active model load.
This value reflects the final layer count passed to native model load after backend policy/fallback decisions, or 0 when no GPU can take them: the registry holds no GPU device and the load lists none. A load that lists a device, such as BLAS, keeps its layers.
Implementation
Future<int?> getResolvedGpuLayers();