PII Arena
OPF detector

opf-ru PII detection benchmark

Measured on 40 of 41 retained datasets under the same frozen scoring and masking protocol.

49.51%fully hidden annotations
66.49%detected / overlapped
9.41%extra masking
2,839CPU characters / second

Strengths

Highest complete-masking rates by normalized data type.

Passwords, keys & tokens85.87%5,733 / 6,676
Logins & usernames85.03%9,414 / 11,071
Phone numbers & email76.35%16,176 / 21,188

Watch list

Lowest observed complete-masking rates; inspect before deployment.

Network identifiers1.96%502 / 25,631
Organizations8.88%1,711 / 19,274
Other sensitive attributes12.35%396 / 3,207

Language coverage

The overall leader can differ from the best choice for one language slice.

LanguageDatasetsFully hidden ↑Detected ↑Extra masking ↓
English1759.39%68.63%10.24%
Russian1843.33%63.41%7.88%
Multilingual557.79%72.57%0.62%

Configuration, source and version

Family
OPF
Revision
ae813a5aaeb31def75e87064818f0b1fbc65e31f
Eligible datasets
40 / 41
Character F1
0.595
Untouched diagnostic
33.51% · 74,621 / 222,661
Benchmark version
1.0.3 · 2026-09-09
CPU seconds / 10k chars
3.522
GPU seconds / 10k chars
1.118

source overlap: russian-pii-66k

Results by dataset

Complete masking is primary; detection shows whether any character overlapped.

DatasetGold annotationsFully hidden ↑Detected ↑Extra masking ↓
rubai-ru3,4585.64%97.60%
multiconer-ru1,2086.13%20.36%3.07%
tab-echr3,83019.61%27.05%
alexen21,26123.47%95.72%11.46%
nerel24,36929.58%37.72%
synth-ru-tickets32,09934.56%56.78%
factrueval7,96636.35%38.24%
corrupt-redmadrobot5,53142.27%65.12%5.95%
dialogpii-multi10,34544.62%59.58%
dialogpii-en3,08745.19%56.30%10.26%
alrosait1,86245.60%98.82%5.41%
jayguard1,19549.04%73.22%3.32%
synth-wiki-tables13,09449.77%75.36%
ameau012,24250.22%60.17%3.25%
synth-jira-comments2,70551.68%68.39%
gretel-multi4,48451.90%74.11%0.55%
redmadrobot5,51652.83%71.57%1.85%
scanpatch8,70854.67%76.53%2.73%
redact-multi22,86458.26%74.80%
tonicai2,41759.45%63.14%0.27%
privy1,89959.87%74.72%11.62%
nemotron-pii9,39162.40%72.02%1.76%
redact-ru9,23063.13%80.44%8.88%
corrupt-hivetrace1,66763.53%91.24%9.48%
nym-en3,58465.71%77.43%4.45%
kiji-multi7,56566.72%74.63%
synth-env-configs39267.35%99.49%23.09%
nym-multi8,16968.10%80.01%5.24%
kiji-en7,62369.15%75.30%
corrupt-secrets-issues28870.83%97.57%8.62%
secrets-issues28872.92%98.26%5.85%
nym-ru9,49775.66%88.46%
leak-museum10177.23%85.15%4.79%
synth-secrets-en58177.45%98.28%29.14%
synth-secrets-ru58179.69%99.83%31.14%
leaky-repo9580.00%91.58%3.32%
hivetrace1,66782.00%95.02%1.42%
creddata77692.65%97.04%4.90%
secrets-rules74698.39%99.60%49.17%
arthur-passwords28098.93%98.93%13.19%