Glossary
Benchmark
A standard test used to compare models, such as a set of coding tasks or exam questions.
See How to read an AI benchmark. Vendor-run results are claims; independent runs are evidence.
Related terms
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.