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

Hallucination Evaluation research benchmarks – Page 3 · SOTA2 Research

Hallucination Evaluation

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

POPE v1.0 (test)

87POPE Score

15

Mar 24, 2026

POPE MSCOCO, A-OKVQA, GQA average (Adversarial)

84.11Accuracy

15

Feb 26, 2026

POPE Popular MSCOCO, A-OKVQA, GQA average

88.48Accuracy

15

Feb 26, 2026

POPE MSCOCO, A-OKVQA, GQA average (Random)

93.24Accuracy

15

Feb 26, 2026

EventHallusion binary QA (test)

0.655Accuracy

15

Feb 26, 2026

VRIPT-HAL (test)

52.9F1 Score

15

Feb 26, 2026

75.6Score

14

May 28, 2026

MCEval HaluEval 8K (test)

82.2Accuracy

14

Feb 26, 2026

44.6Score

13

Jul 1, 2026

70.24Score

13

May 19, 2026

POPE COCO 2014 (Adversarial)

81.43Accuracy

13

Feb 26, 2026

POPE COCO 2014 (Random)

90.1Accuracy

13

Feb 26, 2026

89.9Accuracy

13

Feb 26, 2026

HallusionBench 2024

52.2Score

13

Feb 26, 2026

ThinkLite-VL-7B

72.97Score

12

Apr 27, 2026

Knowledge Deficiency Hallucination Open-ended Evaluation (test)

92.75BertScore

12

Apr 3, 2026

HallucinationBench

75Score

12

Feb 26, 2026

89.87Accuracy

11

Jul 3, 2026

66.6REG

11

Mar 10, 2026

HallusionBench GPT4-assisted (All)

49.94Accuracy (All)

11

Feb 26, 2026

HallusionBench 1.0 (test)

22.2fACC

10

May 13, 2026

94.85Recall (Positive)

10

Apr 24, 2026

78.51CPQ

10

Apr 24, 2026

TruthfulQA Estonian

Llama-3.1-8B-Instruct

43.7Accuracy

10

Mar 4, 2026

68.73MiniGPT-4 Score

10

Feb 26, 2026

Page 3 of 6

Popular datasets

POPE MME CHAIR Style-CHAIR HallusionBench

Follow for updates

Join SOTA2 Contact