Harvey LAB-AA - White Collar Defense & Investigations: leaderboard
Metric: Criterion Pass Rate (%). Source: artificialanalysis.ai. 43 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Muse Spark 1.3 (xHigh) | 97.31 |
| 2 | Claude Opus 5 (Adaptive Reasoning, Max Effort) | 94.61 |
| 3 | Kimi K3 (Max) | 93.27 |
| 4 | Grok 4.5 (High) | 91.25 |
| 5 | Claude Sonnet 5 (Adaptive Reasoning, Max Effort) | 91.25 |
| 6 | GLM-5.2 (Max) | 90.57 |
| 7 | MiniMax-M3 | 89.56 |
| 8 | Gemini 3.7 Flash (High) | 88.22 |
| 9 | Gemini 3.6 Flash (High) | 85.52 |
| 10 | Claude Opus 4.8 (Adaptive Reasoning, Max Effort) | 85.52 |
| 11 | GPT-5.6 Luna (Max) | 85.19 |
| 12 | Kimi K2.6 | 82.83 |
| 13 | Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort) | 81.82 |
| 14 | Qwen 3.7 Max | 81.14 |
| 15 | GPT-5.6 Sol (Max) | 80.81 |
Interactive version: theaggregate.ai/benchmark?slug=harvey-lab-aa-white-collar-defense-investigations · How It Works · Data refreshed daily, snapshot 2026-09-19.