Scoring & evidence
Leaderboard
A ranked table of systems evaluated under stated conditions.
Definition
A leaderboard orders models or systems using one or more benchmark scores and usually reports the configuration and submission date.
Why it matters
It is a useful discovery tool, not a procurement decision: scores may use different budgets, harnesses or model versions.
Related concepts
- Benchmark
A standardised test used to compare systems consistently.
- System under test
The complete configuration being evaluated—not only the model.
- Confidence interval
A range expressing uncertainty around an estimated score.