TriBench-Ko - Adjudicative Overreach: leaderboard

The adjudicative overreach risk dimension of TriBench-Ko (2026), a Korean benchmark of attorney-reviewed yes/no items built from court decisions, where false statements carry statements that presuppose judicial authority reserved to the judge. The score is macro-F1 on a 0-100 scale over the atomic binary judgments, zero-shot at temperature 0; higher is better.

Metric: Macro-F1 (0-100) on the TriBench-Ko Adjudicative Overreach risk items (statements that presuppose judicial authority reserved to the judge): yes/no verification of statements about Korean court decisions, computed over the atomic binary judgments, zero-shot, temperature 0, at most 64 output tokens; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 13 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Qwen 3.5 9B88.2#828
2GPT-5.487.7#135
3Midm-2.0-Base-Instruct84.4#1988
4GPT-5.4 Mini82.9#406
5Ministral-3-8B-Instruct-251279.4#1907
6Phi-474.6#1931
7Gemma 3 12B (IT)71.6#1887
8GPT-4o66.4#751
9kanana-1.5-8B-instruct-250561.2#3315
10Qwen 3 8B50.5#1873
11EXAONE 3.5 7.8B Instruct46.2#3410
12Llama 3.1 8B Instruct41.2#4600
13A.X-3.1-Light37.5

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=tribench-ko-adjudicative-overreach · How It Works · Data refreshed daily, snapshot 2026-11-21.