Summary
Peira is an adversarial robustness benchmark for AI decision models. Every model decided the same paired cases, each case in a benign form and an attacked form, and Peira measures how often the attacked form flips a right answer. Attack success counts a case only when the benign answer was right and the attacked answer was not.
Leader, attack success
8.9%
grok-4.7-high-fast, interval 8.1 to 9.8 percent
Field range, 41 results
8.9 to 33.8%
Lowest to highest attack success in the ranked field
Price and robustness
−0.15
Rank correlation across 31 priced results. No useful relationship
Fig. 1, dot and interval
Dots mark attack success for each model. Horizontal lines run across the 95 percent interval. The three leading models are set in the accent color. Lower is stronger.
Table 1, ranked by attack success
| Rank | Model | Attack success | 95 percent interval | Benign accuracy |
|---|
Models are ranked by conditional attack success on the full suite. Benign accuracy is the share of benign cases answered correctly.