TriBench-Ko: leaderboard
Large language models (LLMs) are increasingly integrated into legal workflows. However, existing benchmarks primarily address proxy tasks, such as bar examination performance or classification, which fail to capture the performance and risks inherent in day-to-day judicial processes. To address this, we publicly release TriBench-Ko, a Korean benchmark designed to evaluate potential deployment risk
Metric: Macro F1 (self-reported). Source: benchmarklist.com. Saturation forecast: Around May 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 | 83.5 | #79 |
| 2 | GPT-5.4 Mini | 78.1 | #212 |
| 3 | Qwen 3.5 9B | 77.1 | #379 |
| 4 | GPT-4o | 71.5 | #343 |
| 5 | Phi-4 | 71.2 | #706 |
| 6 | Gemma 3 12B (IT) | 62.9 | #664 |
| 7 | Qwen 3 8B | 60.8 | #674 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=tribench-ko · How It Works · Data refreshed daily, snapshot 2026-10-20.