##### ###### ##### ### # # ### # # ###### ## ## ## ## ## ## ## # # # # # ## ##### #### ##### # # # # # # # #### ## # ## ## ## ## # # # # # ## ## # ###### ## ### # ### # ######
##### ###### ##### ### # # ### # # ###### ## ## ## ## ## ## ## # # # # # ## ##### #### ##### # # # # # # # #### ## # ## ## ## ## # # # # # ## ## # ###### ## ### # ### # ######
What's ahead
So far, you've run one agent on one GPU. That gets you about experiments per hour.
This section shows you what happens when you give the agent GPUs, how agents coordinate through AgentHub, and how to run AutoResearch on consumer hardware. You'll also learn when scaling helps and when it wastes money.
