Interactive partially observable Markov decision processes (I-POMDP) provide\na formal framework for planning for a self-interested agent in multiagent\nsettings. An agent operating in a multiagent environment must deliberate about\nthe actions that other agents may take and the effect these actions have on the\nenvironment and the rewards it receives. Traditional I-POMDPs model this\ndependence on the actions of other agents using joint action and model spaces.\nTherefore, the solution complexity grows exponentially with the number of\nagents thereby complicating scalability. In this paper, we model and extend\nanonymity and context-specific independence -- problem structures often present\nin agent populations -- for computational gain. We empirically demonstrate the\nefficiency from exploiting these problem structures by solving a new multiagent\nproblem involving more than 1,000 agents.\n