Evaluation & Quality
Benchmark
A standardized test or dataset used to compare model or system performance.
Example
Several models answer the same public test so their results can be compared.
Why it matters
It gives teams a common reference for comparing systems under stated conditions.
You'll hear it
“That score comes from a test unlike our everyday requests.”
Can you control it?
Set by the team
The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.
Common questions
- Can a system do well on a benchmark but poorly at work?
- Yes. The test may leave out important tasks, users or operating conditions.
- Can public test questions become familiar to an AI?
- Yes. Exposure during development can make the results less independent.
- Does one benchmark measure every useful ability?
- No. Different tests emphasize different tasks and can overlook qualities such as cost or ease of use.