On an agent’s configuration screen, Pro, Scale, and Team plan members may adjust the intelligence of the LLM used for processing their agent’s tasks. Higher intelligence selections will boost the agent’s ability to perform complex tasks, at the cost of higher task credit usage. We recommend you experiment with the setting on agents and use the lowest intelligence model that provides satisfactory results.
Complex tasks often require advanced reasoning to produce satisfactory results. In these scenarios, the increasing intelligence will produce acceptable results in a quicker fashion. While task credits will be used more quickly, this result is often achieved with less back-and-forth with the agent, leading to more task credit efficiency in the long run.
Agents performing less complex tasks will often perform just as well with a lower intelligence level selected. Opting to use a lower intelligence level will be for task credit, and thus cost-efficient.
