Re-run logs with different prompts or models

This can be helpful to spot check to see if changing the prompt or model would improve the response.