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

Language Identification research benchmarks – Page 4 · SOTA2 Research

Language Identification

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

Chinese dialects

88.47Accuracy

2

Mar 12, 2026

UDHR 142 langs (out-of-domain)

86.8Macro F1

2

Feb 26, 2026

Tatoeba 201 langs (out-of-domain)

41.4Macro F1

2

Feb 26, 2026

FastSpell Nynorsk (re-annotated)

79.07NB F1 Score

2

Feb 26, 2026

FLORES-200 ∩ NLLB 193 languages (test)

fasttext LID model

95.9F1 Score

2

Feb 26, 2026

98.7Accuracy

2

Feb 26, 2026

93Accuracy

2

Feb 26, 2026

TCL UTF-8 converted

0.947Accuracy

2

Feb 26, 2026

Ensemble (GlotLID & OpenLID-v3)

99.99Loose Accuracy

1

Feb 26, 2026

Ensemble (GlotLID & OpenLID-v3)

1Loose Accuracy

1

Feb 26, 2026

Nordic DSL (test)

Ensemble (GlotLID & OpenLID-v3)

0.937Loose Accuracy

1

Feb 26, 2026

Ensemble (GlotLID & OpenLID-v3)

94.62Loose Accuracy

1

Feb 26, 2026

Bhasha-Abhijnaanam 24 languages native-script (test)

98.11Precision

1

Feb 26, 2026

FLORES-200 201 languages (test)

fasttext LID model

92.7F1

1

Feb 26, 2026

Page 4 of 4

Popular datasets

FLORES UDHR FLEURS ML-SUPERB GlotLID-C

Follow for updates

Join SOTA2 Contact