Clean versus distorted performance

The table below is the NTIRE 2026 public-test leaderboard from Table 3 of Gushchin et al.. The published entries are 7 billion parameter models. Our model, Seer, at 302 million parameters sits third on robust ROC AUC, state of the art at this scale.

MethodParamsROC AUCRob. ROC AUC
MICV7B0.99780.9738
Ant International7B0.99730.9731
Seer (ours)302M0.96770.9228
TeleAI-TeleGuard7B0.97620.9215
INTSIG7B0.98100.9090
vincentlc7B0.94970.8633
UESTC7B0.96930.8558
Reagvis Labs7B0.94230.8474
PSU7B0.91320.8334
Shallow Real7B0.99540.8302

Published scores from Table 3 of Gushchin et al., NTIRE 2026 Challenge on Robust AI-Generated Image Detection in the Wild