SAC (Soft Actor-Critic) is a popular off-policy deep reinforcement learning algorithm that optimizes both expected return and policy entropy to achieve stable and efficient learning in continuous control tasks.