LexKairos - Action Limitation Reasoning: leaderboard

Metric: F1 score (%; 109 items: decide whether a limitation period for an action has expired; 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 December 2026. 8 models tracked.

Top models

#ModelScore
1Gemini 3 Flash (Minimal)83.02
2Qwen 3.5 9B (Non-reasoning)77.07
3DeepSeek R168.33
4DeepSeek V4 Flash (Non-reasoning)67.74
5GPT-5.4 (Non-reasoning)60.52
6Llama 3.1 8B8.34

Interactive version: theaggregate.ai/benchmark?slug=lexkairos-action-limitation-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-29.