Harvey HLAB - Criteria Pass Rate - Structured Finance Securitization
Professional Work · 2026-05-06
The Structured Finance Securitization 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 1.3 | 91.1 |
| Muse Wonder Full | 91.1 |
| Grok 4.6 | 90.8 |
| Muse Wonder | 90.5 |
| Muse Spark 1.2 | 88.9 |
| Grok 4.7 | 88.6 |
| Gemini 4 Argon | 88.3 |
| Muse Spark 1.1 | 87.3 |
| Gemini 3.7 Flash | 87.0 |
| GLM-5.3 | 87.0 |
| DeepSeek V4.1 Flash 1017 | 86.7 |
| MiMo-V2.6-Pro | 86.7 |
| Hy4 Preview | 86.3 |
| GLM-5.3 Flash | 86.0 |
| Qwen3.8-Max | 86.0 |
| Ember-1 | 85.7 |
| Gemini 3.6 Flash | 85.7 |
| Kimi K3 | 85.7 |
| DeepSeek V4 Pro 0813 | 85.4 |
| Gemini 3.8 Flash | 85.4 |
| MiMo-V2.6-Flash | 85.4 |
| Grok 4.5 | 85.1 |
| Claude Opus 5 | 84.8 |
| DeepSeek-V4-Flash-0731 | 84.8 |
| Step 5 Preview | 84.4 |
| GPT-6.1 Sol | 83.8 |
| GPT-6 Astra | 83.8 |
| GLM-5.2 | 83.5 |
| DeepSeek V4.1 Flash | 82.5 |
| Claude Haiku 5.5 | 82.3 |
| DeepSeek-V4-Pro | 82.2 |
| GPT-5.6 Luna | 81.6 |
| Opus 4.8 | 81.3 |
| MiniMax M3 | 81.3 |
| Opus 4.7 | 81.0 |
| Sonnet 4.6 | 81.0 |
| GPT-6 Luna | 81.0 |
| MiMo-V2.5-Pro | 81.0 |
| Claude Opus 5.5 | 79.9 |
| GPT-5.6 Sol | 79.4 |