LegalBench - Issue Tasks
Professional Work · 2023-08-18
The Issue Tasks split of LegalBench. This child benchmark separates a source-reported subtask or subtrack from the parent aggregate so scores at different grains do not share one benchmark_slug.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 4 Argon | 92.9 |
| Claude Fable 5.1 | 92.7 |
| Gemini 3.7 Flash | 92.1 |
| Claude Fable 5 | 92.1 |
| Claude Opus 5 | 91.9 |
| Gemini 3.1 Pro Preview | 91.5 |
| GPT-5.6 Sol | 91.4 |
| Gemini 3.8 Flash | 91.3 |
| GPT-5.5 | 91.2 |
| Gemini 3 Pro Preview | 91.2 |
| Qwen3.5 Plus (2026-02-15) | 91.2 |
| Gemini 3 Flash Preview | 91.1 |
| Gemini 3.6 Flash | 91.1 |
| Muse Wonder Full | 91.0 |
| Kimi K3 | 90.9 |
| Qwen3.5-Flash (source unspecified) | 90.9 |
| MiniMax M3 | 90.9 |
| Opus 4.6 | 90.7 |
| Qwen3.8-Max | 90.6 |
| GPT-5.4 | 90.5 |
| GLM-5.3 | 90.4 |
| Qwen3.8 27B | 90.2 |
| GLM-5.3 Flash | 90.1 |
| Grok 4.5 | 90.1 |
| Grok 4.6 | 90.1 |
| Kimi K2.6 | 90.1 |
| Hy4 Preview | 90.0 |
| GPT-5.1 | 90.0 |
| Qwen3.7-Max (source unspecified) | 89.7 |
| Qwen3.6 Plus (2026-04-02) | 89.7 |
| GPT-5 | 89.5 |
| GLM-5.2 | 89.5 |
| Grok 4.7 | 89.5 |
| Gemini 2.5 Flash Preview 04-17 | 89.5 |
| Grok 4 | 89.4 |
| GPT-5.6 Terra | 89.4 |
| Muse Spark 1.2 | 89.4 |
| Gemini 3.1 Flash-Lite Preview | 89.4 |
| Gemini 3.5 Flash-Lite | 89.3 |
| Grok 4.1 Fast | 89.3 |