Legal Research Bench Weighted Pass Administrative/Regulatory
Professional Work · 2026-06-23
Weighted-pass practice-area score for Administrative/Regulatory tasks in Vals AI Legal Research Bench.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5 | 92.7 |
| Kimi K3 | 90.7 |
| GPT-5.6 Sol | 90.7 |
| Claude Fable 5 | 90.2 |
| Grok 4.5 | 88.8 |
| Sonnet 5 | 87.8 |
| GPT-5.6 Luna | 87.5 |
| Opus 4.8 | 87.4 |
| Gemini 3.5 Flash | 87.1 |
| Sonnet 4.6 | 86.8 |
| GPT-5.6 Terra | 86.7 |
| GPT-5.5 | 86.7 |
| MiniMax M3 | 86.2 |
| GLM-5.2 | 86.0 |
| Inkling | 84.1 |
| Qwen3.7-Max | 84.0 |
| Muse Spark 1.1 | 83.6 |
| Gemini 3.6 Flash | 83.5 |
| DeepSeek-V4-Pro | 83.4 |
| Inkling-Small | 83.0 |
| GLM-5.1 | 80.0 |
| Gemini 3.1 Pro Preview | 79.2 |
| Kimi K2.6 | 75.5 |
| Grok 4.3 | 73.9 |
| Gemini 3.5 Flash-Lite | 72.0 |
| GPT-5.4 Mini | 69.7 |
| Laguna M.1 | 66.6 |
| Laguna XS.2 | 58.1 |