Legal Research Bench Family
Professional Work · 2026-06-23
This row reports the practice-area score for Family tasks in Vals AI Legal Research Bench.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 4 Carbon | 45.5 |
| Claude Opus 5 | 40.9 |
| Muse Spark 1.3 | 40.9 |
| Gemini 4 Argon | 36.4 |
| Muse Wonder | 36.4 |
| Qwen3.8 Max (0902) | 36.4 |
| Opus 4.7 | 31.8 |
| GLM-5.3 | 31.8 |
| Hy4 Preview | 31.8 |
| MiMo-V2.6-Flash | 31.8 |
| Muse Wonder Full | 31.8 |
| Sonnet 4.6 | 27.3 |
| DeepSeek V4.1 Flash 1017 | 27.3 |
| Ember-1 | 27.3 |
| GLM-5.3 Flash | 27.3 |
| MiMo-V2.6-Pro | 27.3 |
| Muse Spark 1.2 | 27.3 |
| Claude Haiku 5.5 | 22.7 |
| DeepSeek V4 Pro 0813 | 22.7 |
| GLM-5.2 | 22.7 |
| Grok 4.6 | 22.7 |
| Kimi K3 | 22.7 |
| Muse Spark 1.1 | 22.7 |
| Muse Spark Open | 22.7 |
| Qwen3.8 27B | 22.7 |
| Beam 1.0 501B A23B | 18.2 |
| Opus 4.8 | 18.2 |
| Sonnet 5 | 18.2 |
| DeepSeek-V4-Flash-0731 | 18.2 |
| Gemini 3.5 Flash | 18.2 |
| Gemini 3.8 Flash | 18.2 |
| GPT-5.6 Luna | 18.2 |
| GPT-5.6 Terra | 18.2 |
| GPT-6.1 Sol | 18.2 |
| GPT-6 Astra | 18.2 |
| Grok 4.7 | 18.2 |
| Qwen3.8-Max | 18.2 |
| DeepSeek V4.1 Flash | 13.6 |
| Gemini 3.7 Flash | 13.6 |
| GPT-5.5 | 13.6 |