Harvey HLAB - Criteria Pass Rate - Litigation Dispute Resolution
Professional Work · 2026-05-06
The Litigation Dispute Resolution 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 |
|---|---|
| Muse Spark 1.2 | 95.1 |
| Muse Spark 1.3 | 94.4 |
| Grok 4.6 | 94.0 |
| Muse Spark 1.1 | 93.6 |
| Ember-1 | 92.9 |
| Gemini 4 Argon | 92.9 |
| Muse Wonder Full | 92.5 |
| Kimi K3 | 91.8 |
| Grok 4.7 | 91.4 |
| Grok 4.5 | 90.6 |
| Muse Wonder | 90.6 |
| Step 5 Preview | 90.6 |
| MiMo-V2.6-Flash | 90.3 |
| Qwen3.8-Max | 90.3 |
| Opus 4.8 | 89.9 |
| Gemini 3.8 Flash | 89.9 |
| GLM-5.3 | 89.9 |
| Sonnet 4.6 | 89.5 |
| Hy4 Preview | 89.5 |
| Mistral Large 4 Preview | 89.5 |
| Qwen3.8 27B | 89.5 |
| Gemini 3.7 Flash | 88.8 |
| MiMo-V2.6-Pro | 88.8 |
| MiniMax M3 | 88.8 |
| DeepSeek V4.1 Flash | 88.4 |
| DeepSeek V4.1 Flash 1017 | 87.6 |
| GPT-6 Astra | 87.6 |
| DeepSeek-V4-Flash-0731 | 86.9 |
| GPT-5.6 Luna | 86.9 |
| Qwen3.7-Max (source unspecified) | 86.9 |
| Gemini 3.6 Flash | 86.5 |
| GLM-5.2 | 86.5 |
| GPT-6.1 Sol | 86.5 |
| Claude Opus 5 | 86.1 |
| Opus 4.7 | 85.8 |
| Sonnet 5 | 85.8 |
| Qwen3.6 Plus (2026-04-02) | 85.8 |
| DeepSeek V4 Pro 0813 | 85.4 |
| MiMo-V2.5-Pro | 85.0 |
| DeepSeek-V4-Pro | 84.6 |