Cultural Alignment on CulturalBench Easy
90.8AccuracyCultureBank
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CultureBankFamily=GPT-4.12025.04 | 90.8 | 78.5714 | |
| CAReDiOFamily=GPT-4.12025.04 | 90.32 | 71.4286 | |
| Raw ModelFamily=GPT-4.12025.04 | 89.82 | — | |
| GPT-5 + Role-PlayFamily=Proprietaty LLMs2025.04 | 89.55 | 85.7143 | |
| Role-PlayFamily=GPT-4.12025.04 | 89.29 | 57.1429 | |
| GPT-5Family=Proprietaty LLMs2025.04 | 88.79 | — | |
| CAReDiOFamily=Gemma-3-27B-IT2025.04 | 82.56 | 71.4286 | |
| Raw ModelFamily=Gemma-3-27B-IT2025.04 | 82.11 | — | |
| CultureParkFamily=Gemma-3-27B-IT2025.04 | 81.85 | 64.2857 | |
| CultureDataFamily=Gemma-3-27B-IT2025.04 | 81.83 | 71.4286 | |
| CultureBankFamily=Gemma-3-27B-IT2025.04 | 81.82 | 71.4286 | |
| CultureSPAFamily=Gemma-3-27B-IT2025.04 | 81.4 | 64.2857 | |
| Role-PlayFamily=Gemma-3-27B-IT2025.04 | 81.33 | 64.2857 | |
| CultureLLMFamily=Gemma-3-27B-IT2025.04 | 80.46 | 57.1429 | |
| CultureInstructionFamily=Gemma-3-27B-IT2025.04 | 76.18 | 21.4286 | |
| CAReDiOFamily=Qwen2.5-7B-Instruct2025.04 | 73.48 | 78.5714 | |
| CultureDataFamily=Qwen2.5-7B-Instruct2025.04 | 72.83 | 50 | |
| CultureInstructionFamily=Qwen2.5-7B-Instruct2025.04 | 72.77 | 57.1429 | |
| Role-PlayFamily=Qwen2.5-7B-Instruct2025.04 | 72.38 | 57.1429 | |
| CultureBankFamily=Qwen2.5-7B-Instruct2025.04 | 72.28 | 57.1429 | |
| Raw ModelFamily=Qwen2.5-7B-Instruct2025.04 | 72.01 | — | |
| CultureParkFamily=Qwen2.5-7B-Instruct2025.04 | 71.99 | 50 | |
| CultureLLMFamily=Qwen2.5-7B-Instruct2025.04 | 71.79 | 50 | |
| CultureSPAFamily=Qwen2.5-7B-Instruct2025.04 | 70.92 | 57.1429 |