Claude Opus 4, Claude Sonnet 4
Anthropic
partial detail 17 method fields published
Evaluation
Discriminatory bias evaluation (Claude 4)
open-ended prompts across topics and identity attributes (gender, race, sexual orientation, religion, region) with explicit and inferred attribute specification; plus BBQ benchmark
How it was run
Open a release to see the setup, scoring, and other details that the lab published. Only fields the lab actually disclosed are shown.