A research agent usually finishes in 8 to 12 steps, but a few runs loop 200+ times, re-issuing near-identical searches and running up large bills. Which control most reliably prevents this?

1. A Remove the search tool's ability to return empty results so the model always has new information to reason over each step.
2. B Add a system-prompt instruction telling the model to stop when it believes it has gathered enough information to answer.
3. C Enforce host-side budgets (e.g., 20 steps, token and wall-clock caps) and end with a partial answer or escalation when hit.
4. D Switch to a larger model with a longer context window so it can remember earlier searches and avoid repeating queries.