PsychEthicsBench - Ethical Knowledge: leaderboard
Ethical-knowledge task of PsychEthicsBench (Shen et al., Monash University, 2026): 1,377 single- and multiple-answer multiple-choice questions grounded in Australian psychology and psychiatry codes of ethics and conduct (APS, Ahpra, RANZCP), 63 of them official National Psychology Examination samples and the rest generated by GPT-5 and Claude Sonnet 4.5 under expert-refined prompts. The model answers in the context of Australia; the score is exact match in percent (full credit only for the exact answer set), and higher is better.
Metric: Exact match (%; 1,377 single- and multiple-answer questions asked in the context of Australia; full credit only for the exact answer set). Source: arxiv.org. Saturation forecast: Around February 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 2 13B Chat | 49.02 | #8238 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=psychethicsbench-ethical-knowledge · How It Works · Data refreshed daily, snapshot 2026-11-12.