Glossary
Thinking level
A vendor setting that controls how much reasoning a model does before answering, trading latency and tokens for accuracy; Google's Gemini 4 models offer low, medium, high and xhigh.
The same model can score very differently at different thinking levels, and cost very differently too, because reasoning tokens are billed as output. Gemini 4 Hydrogen reports an Artificial Analysis index of 51 at high and 54 at xhigh. When comparing vendors' benchmark claims, check which level was used; when comparing bills, measure tokens per task at the level you actually run.
Related terms
BenchmarkReasoning budgetReasoning model
Get the brief in your inbox
One email every morning: the stories that mattered in AI, each with a one-line "why it matters" for your role. Free, no ads, unsubscribe any time.
Published daily by AI Briefing. We never share your address.