PreAct-Bench - Daily Life: leaderboard

Metric: Prefix Foresight F1 with the full prefix (PFF(100), binary F1 of the unethical class, 0-1 scaled to %): given an agent trajectory truncated just before its first overt unethical step, the monitor predicts whether the trajectory will turn unethical; paired ethical and unethical trajectories per task; default generation parameters; higher is better; the daily life domain. Source: arxiv.org. Saturation forecast: Around 2030. 18 models tracked.

Top models

#ModelScore
1GPT-5.477.9
2Gemma 3 27B71
3Qwen 3 0.6B (Non-reasoning)66.7
4Gemma 3 4B66.4
5GPT-4o62.1
6Qwen 3 30B A3B 2507 Instruct58
7Qwen 3 0.6B (Thinking)54.5
8Qwen 3 8B (Non-reasoning)50.7
9Qwen 3 8B (Thinking)39.4
10Qwen 3 30B A3B 2507 (Thinking)33.2
11Qwen 3 4B 2507 Instruct26.1
12Qwen 3 4B 2507 (Thinking)20.6
13GPT-OSS-20B14.8
14GPT-OSS-120B14.7

Interactive version: theaggregate.ai/benchmark?slug=preact-bench-daily-life · How It Works · Data refreshed daily, snapshot 2026-09-29.