Legal Research Bench Constitutional/Civil Rights
Professional Work · 2026-06-23
This row reports the practice-area score for Constitutional/Civil Rights tasks in Vals AI Legal Research Bench.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Fable 5 | 50.0 |
| Claude Opus 5 | 44.7 |
| Sonnet 5 | 44.7 |
| Kimi K3 | 44.7 |
| Opus 4.8 | 39.5 |
| GPT-5.6 Terra | 36.8 |
| Sonnet 4.6 | 34.2 |
| GLM-5.1 | 34.2 |
| GPT-5.6 Luna | 34.2 |
| GPT-5.6 Sol | 34.2 |
| MiniMax M3 | 34.2 |
| Grok 4.5 | 31.6 |
| Inkling | 31.6 |
| Gemini 3.5 Flash | 28.9 |
| GLM-5.2 | 28.9 |
| GPT-5.5 | 28.9 |
| Muse Spark 1.1 | 28.9 |
| Qwen3.7-Max | 23.7 |
| Gemini 3.1 Pro Preview | 21.1 |
| Gemini 3.5 Flash-Lite | 21.1 |
| Kimi K2.6 | 21.1 |
| Gemini 3.6 Flash | 18.4 |
| DeepSeek-V4-Pro | 13.2 |
| GPT-5.4 Mini | 13.2 |
| Grok 4.3 | 13.2 |
| Inkling-Small | 10.5 |
| Laguna M.1 | 2.6 |
| Laguna XS.2 | 0.0 |