Tarsier-34B: benchmark results

Provider: Other.

Aggregate Rating 1555 ± 49, rank #1632 of 10916 rated models, from 6 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
TVBench55.5Average Accuracy77.8
INFACT (Faithfulness) - Clean Accuracy56.8Accuracy (0-1, shown times 100) on clean video (Mode I) for 61.5
INFACT (Factuality)55.6Average reliability under induced perturbations (0-1, shown 53.8
INFACT (Factuality) - Clean Accuracy44.1Accuracy (0-1, shown times 100) on clean video (Mode I) for 30.8
INFACT (Faithfulness)57.3Average reliability under induced perturbations (0-1, shown 30.8
FAVOR-Bench30.34Close-ended ALL accuracy (%)20

Interactive version: theaggregate.ai/model?slug=tarsier-34b · How It Works · Data refreshed daily, snapshot 2026-11-08.