Unverified50% confidenceFactExact time
ONGRID's confidence judge checks answer quality without generating any tokens (zero-token review).
1
Sources
50%
Confidence
Long-term
Relevance
10/18/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe benchmark uses a deterministic checker for scoring—answers are either correct or incorrect—and does not use a large model as a judge.77% similarUnverifiedIn actual evaluation, faithfulness is usually automatically determined with LLM-as-a-Judge or NLI-type models, breaking the answer into atomic claims and checking one by one whether each is entailed by the context; correctness relies on human-annotated reference answers or authoritative fact sources.75% similarUnverifiedThe system's prompt requires that all responses must cite sources, and that 'I don't know' must be returned when retrieved results are irrelevant; adversarial testing demonstrated good refusal capability.71% similarUnverified在生成答案后应增加一道忠实度校验（faithfulness pass），将论断与引用块逐字原文比对并给出supported、partial、unsupported、uncited四种裁决70% similarUnverifiedThe value of a test suite lies not in its quantity, but in whether it can continuously provide trustworthy, actionable feedback.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1058220API
curl https://kongchang.com/api/v1/knowledge/claims/1058220MCP
get_claim(id=1058220)