LegalScope - Case Human: leaderboard
LegalScope: 28 model configurations; automatic grading on 861 exam and 276 case prompts, human validation on 80 exam and 10 case prompts with two lawyers. Scores scaled 0-100. August generation logs explicitly distinguish non-thinking and Low/Medium/High settings; unspecified older settings remain unspecified. Later provider defaults do not rewrite the experiment.. Score (%).
Metric: Score (%). Source: raw.githubusercontent.com. Saturation forecast: Around January 2027. 28 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash (Low) | 85.4 | #101 (Gemini 3 Flash) |
| 2 | Gemini 3 Flash | 82.5 | #101 |
| 3 | Gemini 3.5 Flash (Medium) | 81.2 | #61 (Gemini 3.5 Flash) |
| 4 | DeepSeek V4 Pro (Non-reasoning) | 80 | #106 (DeepSeek V4 Pro) |
| 5 | GLM-5.3 (Low) | 80 | #36 (GLM-5.3) |
| 6 | GPT-5.5 (Medium) | 79.6 | #29 (GPT-5.5) |
| 7 | DeepSeek V3.2 | 78.8 | #204 |
| 8 | GPT-5.5 (High) | 77.9 | #29 (GPT-5.5) |
| 9 | GPT-5.4 Mini | 76.7 | #212 |
| 10 | Kimi K2 (Thinking) | 76.7 | #243 (Kimi K2) |
| 11 | Gemini 2.5 Flash | 75.4 | #242 |
| 12 | Grok 4.20 | 75.4 | #172 |
| 13 | Qwen 3.5 9B (Non-reasoning) | 75 | #384 (Qwen 3.5 9B) |
| 14 | Grok 4.5 | 72.9 | #49 |
| 15 | DeepSeek V4 Flash (Non-reasoning) | 72.9 | #139 (DeepSeek V4 Flash) |
No result here: #4 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=legalscope-case-human · How It Works · Data refreshed daily, snapshot 2026-10-23.