LID Benchmark

Language identification benchmark comparing LID models across FLORES+, MADAR, Gherbal-Multi, ATLASIA-LID, WiLI-2018, CommonLID, and Bouquet.

10rows
accuracyprimary metric
2026-05-06sampled

Metadata

ID
lid_benchmark
Category
Language
Release
Unknown

Metrics

Accuracy, Macro F1, Weighted F1, Macro precision, Macro recall, Benchmark count

Latest Results

Snapshot averages public per-model benchmark summary rows across the eight LID benchmarks. Per-benchmark accuracy and F1 values are preserved in result metadata.

Rank Subject Accuracy Model Match Provenance Sampled
1 glotlid 76.13 — Imported 2026-05-06
2 gherbal-v4 75.63 — Imported 2026-05-06
3 openlid-v2 74.43 — Imported 2026-05-06
4 openlid-v1 72.54 — Imported 2026-05-06
5 nllb-lid 64.11 — Imported 2026-05-06
6 hplt-openlid-v3 55.28 — Imported 2026-05-06
7 gherbal-v3 53.13 — Imported 2026-05-06
8 fastlid-176 46.47 — Imported 2026-05-06
9 gherbal-v2 40.74 — Imported 2026-05-06
10 gherbal-v1 32.36 — Imported 2026-05-06