Freshness
Does what it returns match the web as it is now — new pages found, changed pages not served stale, deleted pages not still handed to you? Three independent failure directions; a tool can be good at one and bad at another. Every tool must be tested against the same page or event. Every run is probe-before, the world changes, probe-after; a run whose before-probe comes out the wrong way is VOID, not failed.
What this capability means
Freshness is the capability for a tool’s returned results to stay aligned with the live web as it changes. A tool has it when it can find newly published pages, stop serving stale copies after a known page changes, and stop returning pages that have been removed. Checking it requires the same page or event across runs, with a probe-before/probe-after sequence; if the before-probe is wrong, that run is void rather than a tool failure.
Boundary: It does not judge whether the tool found the right page in the first place or whether its answer is well supported; those are retrieval and answer-generation questions.
Scenarios that test this capability
A scenario is a real-world situation used to test a capability.
Benchmarks that include this capability
A capability has global identity and may be used by more than one benchmark.
Web Search for AI
Which web search or answer API finds the live-web material that actually supports an answer, keeps up when that material changes, and -- where it writes the answer itself -- says only what its own sources support?