Harvey HLAB - Criteria Pass Rate - Insurance
Professional Work · 2026-05-06
The Insurance practice-area split of Harvey's Legal Agent Benchmark held-out evaluation criteria pass rate. This reports pooled rubric criteria passed within that split and is distinct from the all-pass task-score split.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Fable 5 | 93.4 |
| Claude Opus 5 | 91.6 |
| Kimi K3 | 90.1 |
| Grok 4.5 | 89.4 |
| Muse Spark 1.1 | 87.9 |
| Opus 4.8 | 87.2 |
| GLM-5.2 | 87.2 |
| MiniMax M3 | 87.2 |
| Sonnet 5 | 86.8 |
| Gemini 3.6 Flash | 86.4 |
| Sonnet 4.6 | 84.6 |
| GPT-5.6 Sol | 84.2 |
| GPT-5.5 | 82.8 |
| Qwen3.7-Max | 82.1 |
| GPT-5.6 Luna | 81.0 |
| DeepSeek-V4-Pro | 80.6 |
| GLM-5.1 | 80.6 |
| Inkling | 80.6 |
| Kimi K2.6 | 80.2 |
| GPT-5.6 Terra | 78.0 |
| GPT-5.4 | 73.3 |
| Gemini 3.5 Flash-Lite | 67.0 |
| Gemini 3.5 Flash | 63.7 |
| Laguna M.1 | 58.2 |
| Qwen3.7-Plus | 57.5 |
| Grok 4.3 | 48.0 |
| Laguna XS.2 | 44.8 |