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

#ModelScore
1Gemini 3.1 Pro (Preview)77
2Claude Opus 4.775
3Claude Sonnet 4.674
4GPT-5.474
5Qwen 3.5 122B A10B73
6Gemini 3.1 Flash Lite71
7Qwen 3 235B A22B70
8Qwen 3.6 35B A3B68
9DeepSeek V4 Pro67
10GPT-5.4 Mini66
11DeepSeek V4 Flash64

Interactive version: theaggregate.ai/benchmark?slug=benger · How the rankings work · Data refreshed daily, snapshot 2026-07-22.