List your company

ResearchFormerly WizWand

Phishing Detection

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

phishllm-dataset set 2 (test)

87.5Accuracy

13

Mar 4, 2026

HTML-based PI (test)

0.3ASR

12

Feb 26, 2026

Infrastructure Phishing infra_aware

98.6Infra Recall

11

Mar 27, 2026

97.23Accuracy

9

Apr 23, 2026

98.14Accuracy

9

Apr 23, 2026

Balanced 29,142-email corpus Clean detection

96.75F1-score

7

May 13, 2026

96.57Accuracy

5

Apr 23, 2026

Phishing URL dataset (test)

99.819Accuracy

5

Feb 26, 2026

JPCERT/CC Content-Inaccessible (85 campaigns)

98.8Precision

4

Jun 2, 2026

JPCERT/CC Content-Accessible (18 campaigns)

94.7Precision

4

Jun 2, 2026

Pooled Email Dataset (Nigerian, Nazario, Enron, SpamAssassin, and TREC-07)

MultiPhishGuard

99.8Recall

4

May 26, 2026

UCI, OpenPhish, EvilGinx, and GenAI Average (test)

99.78Accuracy

4

May 19, 2026

67.03Minority Class F1 Score

4

Apr 6, 2026

Mendeley and TR-OP averaged

90.57Precision

3

Mar 10, 2026

96.66Accuracy

2

Jun 2, 2026

Large URL corpus (PhishHaven)

CNN + Random Forest

99.35Accuracy

2

Jun 2, 2026

DataPhish 2025 (test)

98.2Precision

2

Apr 28, 2026

Nazario SpamAssassin (test)

99.5Precision

2

Apr 28, 2026

PhishLLM field-study

DynaPhish/PhishLLM

1,029Revealed Count

2

Mar 4, 2026

DynaPhish field-study

1,350Count Revealed

2

Mar 4, 2026

235,795 URLs Proposed

Hybrid MLP + Blacklist

99.24Accuracy

1

Jun 2, 2026

Hybrid DL (CNN + RNN)

98Accuracy

1

Jun 2, 2026

Comparative ML (RF, SVM)

85Accuracy

1

Jun 2, 2026

PhishSnap (test)

76Precision

1

Feb 26, 2026