Legal Research Bench Weighted Pass Criminal
Professional Work · 2026-06-23
Weighted-pass practice-area score for Criminal tasks in Vals AI Legal Research Bench.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5 | 88.1 |
| Opus 4.8 | 87.7 |
| GPT-5.6 Sol | 86.4 |
| Claude Fable 5 | 86.1 |
| Sonnet 4.6 | 85.8 |
| Muse Spark 1.1 | 85.4 |
| Kimi K3 | 84.3 |
| Sonnet 5 | 84.1 |
| GPT-5.6 Terra | 83.4 |
| Grok 4.5 | 82.6 |
| Gemini 3.5 Flash | 82.4 |
| GPT-5.6 Luna | 82.2 |
| GPT-5.5 | 81.8 |
| Gemini 3.6 Flash | 81.0 |
| GLM-5.2 | 80.1 |
| GLM-5.1 | 78.1 |
| MiniMax M3 | 78.1 |
| Inkling-Small | 78.0 |
| Qwen3.7-Max | 77.7 |
| Inkling | 77.4 |
| DeepSeek-V4-Pro | 76.4 |
| Gemini 3.1 Pro Preview | 76.2 |
| Kimi K2.6 | 75.1 |
| Grok 4.3 | 69.3 |
| Gemini 3.5 Flash-Lite | 68.1 |
| GPT-5.4 Mini | 64.4 |
| Laguna M.1 | 61.6 |
| Laguna XS.2 | 56.4 |