PPT-Bench (Epistemic Attack) - Epistemic Destabilization: leaderboard
Metric: Binary capitulation rate (%): share of items scored 1 or 2 over the 23 Type 1 items (epistemic destabilization: Pyrrhonism, Madhyamaka, Academic Skepticism); a gpt-4o pairwise judge compares each single-turn pressured answer (L1) with the model's own unpressured baseline answer (L0) and scores 0 (held), 1 (partial) or 2 (full capitulation); temperature 0, one generation per item; lower is better. Source: arxiv.org. Saturation forecast: Around January 2027. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V3.1 | 30.4 |
| 2 | Nemotron 3 Super | 34.8 |
| 3 | GPT-OSS-20B (Low) | 39.1 |
| 4 | GPT-OSS-20B | 47.8 |
| 5 | Llama 3.3 70B Instruct | 65.2 |
| 6 | Qwen 3 32B | 65.2 |
| 7 | Mistral Small 3.1 | 69.6 |
| 8 | Ministral 8B | 82.6 |
Interactive version: theaggregate.ai/benchmark?slug=ppt-bench-epistemic-attack-epistemic-destabilization · How It Works · Data refreshed daily, snapshot 2026-10-07.