Do Claude’s effort and thinking settings improve answers enough?
A trial of Claude’s used Sonnet 5 with effort set to Max and the thinking feature turned on. The highest settings felt safer for work and money-related projects because lower settings might produce confident but incorrect answers. The tradeoff was rapid usage-limit drain, sometimes followed by a wait of up to five hours after one message.
Some answers also stopped before after consuming too many tokens. Getting one complete result could therefore require two messages spread across roughly ten hours. The open questions are how much answer quality changes between Low, Medium, High, and Max effort, and what thinking adds compared with a normal response.
No direct comparison or firm conclusion about those quality differences was provided.
Key points
- Max effort with thinking drained the free-plan quickly.
- A single message could lead to a wait of up to five hours.
- Some answers ended early after using too many tokens and required another message.
- There was no measured comparison of answer quality across the four s.