Legal Commentary Generation Quality Evaluation on German Civil Code (BGB) Average
4.5Topical Relevance (Human)GPT-4.5-preview
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| GPT-4.5-previewModel=GPT-4.5-preview2026.05 | 4.5 | 4.75 | 5 | 4.5 | 3.75 | 3.25 | 4.5 | 4.25 | 4.25 | 4.25 | |
| GPT-4.1Model=GPT-4.12026.05 | 3.5 | 4.5 | 4.25 | 4.25 | 3.75 | 3.25 | 3.25 | 4.25 | 3 | 3.5 | |
| o3Model=o32026.05 | 3.25 | 4.25 | 5 | 4.75 | 3.25 | 3.25 | 4.75 | 4.5 | 4.75 | 4.75 | |
| GPT-4oModel=GPT-4o2026.05 | 2.75 | 3.5 | 3.5 | 4.25 | 3.5 | 2.75 | 2.5 | 2.75 | 2.75 | 3 |