TW-LegalBench - Constitutional Law: leaderboard
Metric: Constitutional-law questions: accuracy (%) on four-option single-answer questions from Taiwan's 2020-2024 official legal examinations (civil service, judicial, police and professional), zero-shot chain-of-thought, JSON answer (format failures count as wrong), temperature 0 (1 for gpt-5 and gpt-5.2). Source: arxiv.org. Saturation forecast: Around December 2026. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.5 | 89.4 |
| 2 | GPT-5 | 87.6 |
| 3 | GPT-5.2 | 84.8 |
| 4 | GPT-4o (2024-08-06) | 79.9 |
| 5 | Qwen 3 235B A22B | 78.5 |
| 6 | Llama 3.1 405B | 72.2 |
| 7 | GPT-OSS-120B | 67.2 |
| 8 | nemotron-3-nano-30B-a3B | 63.5 |
| 9 | Qwen 2.5 7B | 59.3 |
| 10 | GPT-OSS-20B | 59.1 |
Interactive version: theaggregate.ai/benchmark?slug=tw-legalbench-constitutional-law · How It Works · Data refreshed daily, snapshot 2026-09-29.