Interpretive Canons (BVerfG) - Not Abstract: leaderboard

Metric: Positive-class F1 (%; not abstract (nicht_abstrakt): whether a sentence of the reasons states a concrete rather than abstract assertion, sentence-level binary classification in German Federal Constitutional Court decisions; expert-written prompt with worked examples (the same for every model), reasoning models at temperature 1.0; decision-disjoint held-out test split (60% of each subtask, 1:3 positives to negatives)). Source: arxiv.org. Saturation forecast: Around December 2026. 4 models tracked.

Top models

#ModelScore
1MiniMax-M380.6
2Gemini 3.5 Flash Lite73.4
3DeepSeek V4 Pro73.1
4DeepSeek V4 Flash71.1

Interactive version: theaggregate.ai/benchmark?slug=interpretive-canons-bverfg-not-abstract · How It Works · Data refreshed daily, snapshot 2026-09-26.