LaborBench — leaderboard
Legal RAG and statutory reasoning benchmark over U.S. unemployment-insurance law across multiple states.
Metric: F1 (self-reported). Source: benchmarklist.com. Status: saturation imminent. 5 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude 3.5 Haiku | 68.8 |
| 2 | DeepSeek V3 | 67.4 |
| 3 | GPT-4o Mini | 67 |
| 4 | Llama 3.1 70B Instruct | 64.7 |
| 5 | Gemini 1.5 Flash | 64.4 |
Interactive version: theaggregate.ai/benchmark?slug=laborbench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.