Comment by loclol101

We are actively exploring how frontier agents behave over different task distributions. Whether the tasks are longer horizon, multi-objective or they require more robust control signals and more. Hope this answers your question