Harbor Adapters - LawBench (Terminus-2): leaderboard
LawBench, Chinese legal knowledge and reasoning tasks from 20 task families, here 181 of its 1,000: the mean per-task score in percent under each family's own metric (exact match, F1, ROUGE or rule-based), higher is better. Each model runs in Terminus-2, the Harbor framework's bash-only terminal agent with no native web or image tools, at its provider's default settings, so results can differ from the same model in its vendor's own agent; the score is the mean of three trials (Harbor Adapters large-scale evaluation; Shi, Lin et al., 2026).
Metric: Mean task score (%; 181 tasks, mean of 3 trials). Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 69.7 | #91 |
| 2 | Claude Opus 4.6 | 65.4 | #100 |
| 3 | Gemini 3 Flash (Preview) | 64.8 | #141 |
| 4 | Claude Sonnet 4.6 | 60.8 | #158 |
| 5 | GPT-5.4 (Medium) | 58.6 | #121 (GPT-5.4) |
| 6 | Claude Haiku 4.5 | 53.8 | #568 |
| 7 | GPT-5 Mini | 48.6 | #349 |
| 8 | GPT-5 Nano | 32.1 | #888 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=harbor-adapters-lawbench-terminus-2 · How It Works · Data refreshed daily, snapshot 2026-11-12.