Legal Research Bench Administrative/Regulatory
Professional Work · 2026-06-23
This row reports the practice-area score for Administrative/Regulatory tasks in Vals AI Legal Research Bench.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5 | 65.2 |
| GPT-5.6 Sol | 60.6 |
| Claude Fable 5 | 57.6 |
| GPT-5.6 Terra | 57.6 |
| Sonnet 5 | 54.5 |
| Kimi K3 | 54.5 |
| GPT-5.5 | 53.0 |
| Grok 4.5 | 51.5 |
| GPT-5.6 Luna | 48.5 |
| Opus 4.8 | 43.9 |
| Sonnet 4.6 | 43.9 |
| Gemini 3.5 Flash | 43.9 |
| GLM-5.2 | 39.4 |
| Muse Spark 1.1 | 37.9 |
| MiniMax M3 | 36.4 |
| Qwen3.7-Max | 36.4 |
| Gemini 3.6 Flash | 31.8 |
| GLM-5.1 | 31.8 |
| Inkling | 31.8 |
| DeepSeek-V4-Pro | 28.8 |
| Inkling-Small | 28.8 |
| Grok 4.3 | 25.8 |
| Gemini 3.1 Pro Preview | 24.2 |
| Gemini 3.5 Flash-Lite | 19.7 |
| Kimi K2.6 | 18.2 |
| GPT-5.4 Mini | 15.2 |
| Laguna M.1 | 3.0 |
| Laguna XS.2 | 1.5 |