SAD - Influence (Situating Prompt): leaderboard

Metric: Score (%, higher is better). Source: situational-awareness-dataset.org. 21 models tracked.

Top models

#ModelScore
1O1 Preview (2024-09-12)93.16
2O1 Mini (2024-09-12)88.09
3GPT-4o87.19
4GPT-4 Preview (0125)86.56
5GPT-4 (0613)85.31
6Claude 3.5 Sonnet82.97
7Claude 3 Opus80.78
8Llama 3 70B Chat77.19
9Claude Instant 1.276.25
10Claude 3 Sonnet74.38
11Claude 3 Haiku72.81
12Llama 2 70B71.25
13Claude 2.170.94
14GPT-4 Base69.38
15Llama 2 70B Chat69.06

Interactive version: theaggregate.ai/benchmark?slug=sad-influence-situating-prompt · How It Works · Data refreshed daily, snapshot 2026-09-19.