BenGER — leaderboard
German-law benchmark for subsumption-based legal reasoning, evaluating model answers across Benchathon, ZJS, and doctrinal-principles corpora.
Metric: Aggregate Accuracy (%). Source: arxiv.org. Status: saturation imminent. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 77 |
| 2 | Claude Opus 4.7 | 75 |
| 3 | Claude Sonnet 4.6 | 74 |
| 4 | GPT-5.4 | 74 |
| 5 | Qwen 3.5 122B A10B | 73 |
| 6 | Gemini 3.1 Flash Lite | 71 |
| 7 | Qwen 3 235B A22B | 70 |
| 8 | Qwen 3.6 35B A3B | 68 |
| 9 | DeepSeek V4 Pro | 67 |
| 10 | GPT-5.4 Mini | 66 |
| 11 | DeepSeek V4 Flash | 64 |
Interactive version: theaggregate.ai/benchmark?slug=benger · How the rankings work · Data refreshed daily, snapshot 2026-07-22.