Categories Companies Research

Loading the SOTA2 catalog…

Explore

Categories Companies Research

Contribute

Join SOTA2 Contact

SOTA2

© 2026 SOTA2

Build in Bellevue, USA

WizWand is now SOTA2 Research

Papers Benchmarks Datasets Tasks

Visual Hallucination Evaluation research benchmarks · SOTA2 Research

Visual Hallucination Evaluation

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

Qwen2-VL-72B-Thinking

76.6Accuracy

156

Jul 7, 2026

18.2CHAIR_i

122

Jun 24, 2026

POPE MS-COCO Adversarial sampling (val)

85.48Accuracy

39

Feb 26, 2026

AMBER Discriminative

89Accuracy

18

Jun 24, 2026

AMBER Generative

6.3CHAIRi

18

Jun 24, 2026

88.93F1 Score

14

May 26, 2026

HallusionBench visual questions

65.8Accuracy

10

Feb 26, 2026

POPE MS-COCO Popular sampling (val)

InstructBLIP-14B

82.77Accuracy

6

Feb 26, 2026

POPE MS-COCO Random sampling (val)

InstructBLIP-14B

88.57Accuracy

6

Feb 26, 2026

Popular datasets

POPE HallusionBench AMBER MSCOCO

Follow for updates

Join SOTA2 Contact