LexKairos: leaderboard
LexKairos (Tsinghua University, 2026) evaluates the legal temporal capabilities of LLMs in the Chinese legal context with nine sub-tasks on statutory temporal knowledge, case chronology and statute-case time-limit reasoning, built from real judicial cases and statutes. The score is the mean of the nine sub-task scores (accuracy or F1) in percent in the vanilla setting, where the model answers directly without extra reasoning; higher is better.
Metric: Mean score (%; average of the nine sub-task scores, accuracy or F1, over 3,800 Chinese legal temporal items from real judicial cases and statutes; vanilla setting, the model answers directly without extra reasoning steps (thinking disabled where the model allows it); temperature 0, 8,000 output tokens). Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek R1 | 79.78 | #473 |
| 2 | GPT-5.4 (Non-reasoning) | 77.47 | #123 (GPT-5.4) |
| 3 | Gemini 3 Flash (Minimal) | 76.71 | #164 (Gemini 3 Flash) |
| 4 | DeepSeek V4 Flash (Non-reasoning) | 67.07 | #258 (DeepSeek V4 Flash) |
| 5 | Qwen 3.5 9B (Non-reasoning) | 55.45 | #803 (Qwen 3.5 9B) |
| 6 | Llama 3.1 8B | 28.11 | #4861 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=lexkairos · How It Works · Data refreshed daily, snapshot 2026-11-13.