Unverified50% confidenceBenchmarkExact time
Across six agents and six model-size settings, 29 of 36 submissions passed independent verification.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/5/2026
First Seen
Valid until: 2/3/2027
Sources
Related Entities
Related Claims
UnverifiedThe researchers tested 22 models and 26 configurations, and found that no configuration passed more than 46 of its 168 runs.72% similarUnverifiedThis benchmark uses 10 original short documents of 200 to 250 words each, with 18 planted anomalies, 37 scoring entries in total (18 dirty and 19 clean), testing 13 models with 74 prompts per model.72% similarUnverifiedThe reference distribution of the 40 observations is: 1 recorded evidence, 33 not localized, 0 explicitly missing, and 6 ambiguous.70% similarUnverifiedIn a preregistered study, confidence intervals relying solely on LLM judges failed to cover the full human-annotation ground truth in 18 of 21 task-and-judge combinations.70% similarUnverifiedThe researchers re-analyzed 540 evaluation results and manually annotated 256 numerical claims across 18 submissions.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1138468API
curl https://kongchang.com/api/v1/knowledge/claims/1138468MCP
get_claim(id=1138468)