EvalMORAAL - Pew Global Attitudes: leaderboard
The same direct rating protocol scored against the Pew Global Attitudes survey: 8 moral topics in 39 countries, 312 country-topic pairs; the score is the Pearson correlation.
Metric: Pearson r between direct chain-of-thought ratings and Pew Global Attitudes Survey country averages. Source: arxiv.org. Saturation forecast: Estimated already saturated. 20 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3 Opus (20240229) | 0.89 | #1103 |
| 2 | GPT-4o (2024-05-13) | 0.88 | #735 |
| 3 | Llama 3.3 70B Instruct | 0.88 | #1357 |
| 4 | Qwen 2.5 7B Instruct | 0.87 | #2564 |
| 5 | O1 Preview (2024-09-12) | 0.87 | #391 |
| 6 | Gemini 1.0 Pro | 0.86 | #4043 |
| 7 | Claude 3 Sonnet (20240229) | 0.85 | #1911 |
| 8 | O1 Mini (2024-09-12) | 0.84 | #809 |
| 9 | deepseek-llm-7B-chat | 0.84 | #7056 |
| 10 | GPT-4o Mini (2024-07-18) | 0.83 | #1510 |
| 11 | GPT-4 (0613) | 0.82 | #1214 |
| 12 | Command-R+ (08-2024) | 0.81 | #2551 |
| 13 | Phi-3 Mini 4K Instruct | 0.8 | #6810 |
| 14 | Gemini 2.0 Flash (Preview) | 0.79 | #736 |
| 15 | Mistral Large 2 (Jul) | 0.78 | #1155 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=evalmoraal-pew-global-attitudes · How It Works · Data refreshed daily, snapshot 2026-11-13.