Unverified50% confidenceOpinionExact time
The author calls the above benchmark a "smoke test, not a real-world accuracy claim," with a test sample of only 4 receipts (3 of which were used for benchmark evaluation).
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/18/2026
First Seen
Valid until: 11/1/2026
Sources
Related Entities
Related Claims
UnverifiedRejection tests should be written before positive tests, because positive tests pass on day one, and the scenarios that truly expose problems are those where there is no data yet an answer is forced.67% similarUnverifiedMost benchmarks use accuracy as the core metric, counting both wrong answers and refusals as incorrect and not distinguishing the difference in harm.67% similarUnverifiedRegular expressions that pass initial benchmark testing may still expose problems under more rigorous adversarial probing, meaning passing a benchmark does not equate to true reliability.67% similarUnverifiedThe author himself calls these preliminary results preliminary observations rather than formal benchmark tests, and the test scenarios, sample size, and evaluation criteria have not been made public.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1064402API
curl https://kongchang.com/api/v1/knowledge/claims/1064402MCP
get_claim(id=1064402)