Model effort (reasoning effort)
Model effort, model effort, reasoning effort
A setting for how much a model may think before answering, usually in levels like low, medium and high. More effort means more thinking tokens: often better results on hard tasks, but longer waits and higher cost.
💭 Imagine …
Like asking a colleague: “Quick gut feeling in the hallway” or “Take an afternoon and work it through properly”.
🔎 Example
“Make this email friendlier” needs low effort. “Find the bug in this pricing logic” deserves high effort. Many tools and APIs offer a setting for it; some choose the effort automatically.
🎯 Why it matters
Effort is a dial between quality, speed and cost. More thinking mainly helps with logic and multi-step tasks. It doesn't replace missing knowledge, so still check facts.
🎓 Prefer step by step?
The AI Literacy course: 6 short modules with quizzes, a solid foundation in about 60 minutes.