Comment by bahaAbunojaim

Executing multiple agents on the same model also works.

I find it helpful to even change the persona of the same agent “the prompt” or the model the agent is using. These variations always help but I found having multiple different agents with different LLMs in the backend works better