Prompt Recovery Archive
Search 20 synthetic sample prompt records, inspect response examples, and explore demonstration metrics.
Search archiveExplore a connected demonstration of prompt recovery, human evaluation, model response comparison, and professional report generation. Select a capability below to inspect the working sample interface.
Search 20 synthetic sample prompt records, inspect response examples, and explore demonstration metrics.
Search archiveCompare sample ChatGPT and Claude responses and complete a six-criterion human grading exercise.
Open grading labCreate an ARC2-branded sample analysis report, print or export evaluation data.
Generate reportAll example prompts, model responses, grades, and performance metrics shown in the public demonstration are synthetic illustrations. They are not evidence of live ChatGPT or Claude API execution or independently verified model performance. The public demonstration does not expose ARC2's private evaluation archive.