Skip to content
CDSBench

Which section of the record establishes the seniority (ranking) of the instrument? Answer with the section identifier (e.g. DOC-1#S4).

Fact extractionEvidencedeterministically verifiedvalue extractionbenchmark v0.1
Download PDF
Can you beat the models?

What’s your conclusion?

The documents are below. Commit before the verified outcome and the model answers are shown.

Verified
Claude
GPT
Gemini
Mastermind
Source documents

The record every model received

  • DOC-1 JPY 2050 Notes — Terms and Conditions (excerpt)
  • DOC-2 Determinations Committee notice — Federal Republic of Aurelia