Data as of Aug 25, 2026 · Based on 3,181,687 AI responses across 10,525 prompts · See how Parse measures this
RobustBenchrobustbench.github.ioRobustBench is a standardized benchmark for adversarial robustness that evaluates models under common threat models (L∞ and L2) and corruptions (e.g., CIFAR-10-C, ImageNet-C) using AutoAttack to track real progress. It offers a Model Zoo with 80+ state-of-the-art robust models and leaderboards for CIFAR-10, CIFAR-100, and ImageNet to compare robust accuracy and facilitate downstream use. It also invites external adaptive-attack evaluations to prevent overfitting and reports the best-known robust accuracy with analyses in its accompanying papers.
Sources
arxiv.org shapes more of what AI says about RobustBench than any other source, at 30% of its citations.
adversarial-robustness-toolbox.org · github.com · link.springer.com · machinelearningauthority.com
The market map
AI Red Teaming Services →Where AI ranks RobustBench
