knowngrounds

Check a page

Paste the URL of a page you publish. We pull out the claims a reader could act on, then ask a model about each one — with search available and without.
Options · 290 models available
The agent being measured. Never the same as the grader.
Marks every answer. Sits a tier above the subject and must differ from it.
Ranked by how much it would matter if someone believed the opposite.
The same question asked more than once, because the answer moves. Odd numbers break majority ties; at 1 there is nothing to disagree with.
Snapshots make two runs comparable; live tells you what happens now.
Sets are frozen per page. Bump this only to write fresh questions.

Runs

4 runs · every claim tested against a frozen question set
actionable retrieval-dependent sound not gradeable what do these mean?
openai/gpt-5.6-luna set 3 0 claims 0 checks
On the pageView report →
$0.01
cost
0m51s
openai/gpt-5.6-luna set 3 failed
What happened →
Interrupted — the server restarted while this run was in flight. Run it again.
openai/gpt-5.6-luna set 3 0 claims 0 checks
On the pageView report →
$0.01
cost
4m07s
openai/gpt-5.6-luna set 3 0 claims 0 checks
On the pageView report →
$0.14
cost
2m24s