ResearchBenchmarksMultimodal Claim Verification on AVerImaTeC (dev)Follow92.3Q-EvalVILLAIN33.12448.48763.8579.213Feb 4, 2026Evaluation ResultsMethodMethodLinksQ-EvalEvid-EvalVeracityJustificationVILLAINLeaderboard Name=HUMAN...Leaderboard Name=HUMANE, Evaluation Model=Gemma-3-27B, Conditioned Threshold λ=0.32026.0292.358.364.554.3AIC CTUConditioned Threshold...Conditioned Threshold λ=0.32026.0282.234.737.530.4BaselineConditioned Threshold...Conditioned Threshold λ=0.32026.0248.813.46.65.8ADA-AGGRConditioned Threshold...Conditioned Threshold λ=0.32026.0235.438.645.437.3