A new swarm intelligence-based metaheuristic optimizer, namely Partial Leader Optimizer (PLO), is presented. PLO contains several autonomous agents that represent the solution. The best solution represents collective intelligence, i.e., the leader. PLO has distinct mechanics in finding the acceptable solution during the given iteration. Every agent moves to a specified target in every iteration. Two options can be chosen to determine the target. First, the target is calculated by pushing the virtual best solution away from the corresponding agent. Second, the target is randomly chosen within the solution space. This target selection is conducted stochastically based on the threshold that is set manually before the iteration. Then, several candidates are generated between the target and the agent's current location. The distance between adjacent candidates is the same. The agent moves to the best candidate and updates the best solution. Simulation is implemented to observe and analyze the PLO’s performance. The well-known 23 benchmark functions are used as the optimization problems. In this simulation, PLO is benchmarked with marine predator algorithm (MPA), particle swarm optimization (PSO), average subtraction-based optimizer (ASBO), slime mold algorithm (SMA), and pelican optimization algorithm (POA). The result shows that PLO is competitive compared to these algorithms, especially in solving fixed-dimension multimodal functions. PLO is better than PSO, MPA, SMA, ASBO, and POA in optimizing 22, 19, 18, 9, and 20 functions out of 23, respectively.