Time-R1-7B: benchmark results

Provider: Other.

Aggregate Rating 1590 ± 52, rank #1042 of 10552 rated models, from 13 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
EG-VQA - Evidence Grounding16.46Evidence-grounded F1 (%, EG-F1 at temporal IoU 0.3 and seman87.5
MMVIAD - Defect Classification31.3Accuracy (%; MMVIAD-Standard test split, 1,101 rendered 2-se60
EG-VQA - Counterfactual21.15Strict accuracy (%) on counterfactual questions (outcomes of50
MMVIAD42.1Mean of three accuracies and visible-time mIoU (%; MMVIAD-St50
MMVIAD - Unseen Categories46.2Mean of three accuracies and visible-time mIoU (%; MMVIAD-Un50
MMVIAD - Anomaly Visible-Time Localization31.1Mean temporal IoU (%; MMVIAD-Standard test split, predicted 45
MMVIAD - Anomaly Detection40.8Accuracy (%; MMVIAD-Standard test split, 1,101 rendered 2-se40
MMVIAD - Object Classification65Accuracy (%; MMVIAD-Standard test split, 1,101 rendered 2-se40
EG-VQA15.66Strict accuracy (%): share of answers the judge rates fully 37.5
EG-VQA - Descriptive10.52Strict accuracy (%) on descriptive questions (directly obser37.5
EG-VQA - Relaxed Accuracy30.44Relaxed accuracy (%): fully correct answers count 1 and part37.5
EG-VQA - Temporal7.74Strict accuracy (%) on temporal questions (event order and p37.5

Interactive version: theaggregate.ai/model?slug=time-r1-7b · How It Works · Data refreshed daily, snapshot 2026-11-16.