LegalScope - Argument Validity: 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 | GPT-5.5 (High) | 94.1 | #29 (GPT-5.5) |
| 2 | GPT-5.5 (Medium) | 92.7 | #29 (GPT-5.5) |
| 3 | GPT-5.4 Mini | 84.9 | #212 |
| 4 | GLM-5.3 (Low) | 83.2 | #36 (GLM-5.3) |
| 5 | Grok 4.5 | 81.3 | #49 |
| 6 | DeepSeek V4 Pro (Non-reasoning) | 78.9 | #106 (DeepSeek V4 Pro) |
| 7 | Qwen 3 Max | 74.5 | #210 |
| 8 | Qwen 3 Max (Thinking) | 74.2 | #210 (Qwen 3 Max) |
| 9 | Gemini 3.5 Flash (Medium) | 72.6 | #61 (Gemini 3.5 Flash) |
| 10 | DeepSeek V3.2 | 72.4 | #204 |
| 11 | Kimi K2 (Thinking) | 72.1 | #243 (Kimi K2) |
| 12 | Gemini 2.5 Flash | 72 | #242 |
| 13 | Gemini 3 Flash (Low) | 71.7 | #101 (Gemini 3 Flash) |
| 14 | Qwen 3.8 27B (Non-reasoning) | 71.6 | #118 (Qwen 3.8 27B) |
| 15 | Gemini 3 Flash | 71.3 | #101 |
No result here: #4 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=legalscope-argument-validity · How It Works · Data refreshed daily, snapshot 2026-10-23.