Effective energy management in battery electric vehicle (BEV) is required to extend battery life, maximize driving range, and minimize energy consumption. Reinforcement learning (RL) is a model-free and very effective approach to learn adaptive control policies in a variable and uncertain driving environment. This work compares Soft Actor-Critic (SAC) and Proximal Policy Optimization (PPO) for energy management in BEVs. A simulated electric powertrain and battery model is developed and tested in a custom simulation environment and various driving cycles to evaluate the robustness of each algorithm. These RL agents are trained to keep SOC at optimal level and minimize energy consumption. The analysis shows that PPO adopts a more conservative and consistent SOC control strategy, resulting in lower energy consumption. However, it generally falls behind SAC in terms of overall reward performance.