Aspects presented herein may enable a first UE to utilize inference results provided by at least a second UE to improve its inference accuracy. In one aspect, a first UE establishes an ML inference result sharing session with one or more second UEs for at least one ML inference task. The first UE receives a first set of ML inference results associated with the at least one ML inference task, where the first set of ML inference results is received from the one or more second UEs during the ML inference result sharing session. The first UE estimates an aggregated ML inference result based on at least one of the first set of ML inference results or a second set of ML inference results, where the second set of ML inference results is configured by the first UE for the at least one ML inference task.