When the result isn’t good enough

Before switching to a more expensive model, check:

Is the request clear?

Does the AI know exactly what to do?

Is the context sufficient?

Does the AI have the documents and data it needs?

Has the output been specified?

Does the AI know what form of result you want?

Is the workflow too long?

If a task is too big, split it into:

Research → Analysis → Critique → Refinement

Is the model suitable?

A more powerful model is not necessarily better if it is a poor fit for the task.

Is the reasoning level high enough?

For complex tasks, try increasing the reasoning level from Medium to High before changing the rest of the setup.