PII Arena
OPF detector

apararti PII detection benchmark

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

62.49%fully hidden annotations
70.23%detected / overlapped
11.16%extra masking
2,446CPU characters / second

Strengths

Highest complete-masking rates by normalized data type.

Dates & times92.44%5,086 / 5,502
Passwords, keys & tokens89.42%6,143 / 6,870
Phone numbers & email89.21%19,610 / 21,981

Watch list

Lowest observed complete-masking rates; inspect before deployment.

Organizations8.89%1,713 / 19,274
Other sensitive attributes15.53%498 / 3,207
Addresses & locations41.87%16,583 / 39,602

Language coverage

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

LanguageDatasetsFully hidden ↑Detected ↑Extra masking ↓
English1769.08%72.31%11.73%
Russian1960.43%70.32%10.25%
Multilingual563.09%68.55%1.57%

Configuration, source and version

Family
OPF
Revision
a1650a1927ed8aceb9e15fa8ed13ffdf60cbf107
Eligible datasets
41 / 41
Character F1
0.658
Untouched diagnostic
29.77% · 67,711 / 227,466
Benchmark version
1.0.3 · 2026-09-09
CPU seconds / 10k chars
4.088
GPU seconds / 10k chars
1.685

Results by dataset

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

DatasetGold annotationsFully hidden ↑Detected ↑Extra masking ↓
multiconer-ru1,2084.72%8.86%3.48%
rubai-ru3,4585.81%94.94%
nerel24,36919.49%22.89%
alexen21,26122.05%92.62%11.20%
factrueval7,96625.78%26.54%
tab-echr3,83026.37%28.85%
dialogpii-multi10,34530.81%38.23%
jayguard1,19548.54%61.67%3.09%
dialogpii-en3,08751.96%56.79%9.13%
ameau012,24257.76%59.95%3.96%
tonicai2,41759.33%62.72%0.36%
corrupt-redmadrobot5,53160.91%72.92%3.62%
synth-jira-comments2,70563.59%75.12%
gretel-multi4,48465.66%73.19%1.54%
alrosait1,86268.15%89.63%6.25%
redact-multi22,86468.32%74.10%
privy1,89969.25%75.57%10.42%
leaky-repo9570.53%84.21%4.00%
redmadrobot5,51671.23%78.21%2.48%
nemotron-pii9,39172.40%74.90%2.55%
corrupt-secrets-issues28872.92%97.22%7.81%
leak-museum10174.26%75.25%3.59%
scanpatch8,70874.44%81.14%2.32%
secrets-issues28874.65%97.92%6.67%
kiji-multi7,56575.97%77.75%
nym-multi8,16976.02%80.34%3.93%
synth-ru-tickets32,09976.16%86.81%
redact-ru9,23077.54%83.15%12.88%
synth-env-configs39278.32%99.23%21.29%
corrupt-hivetrace1,66780.38%90.16%4.87%
nym-ru9,49781.98%88.45%
synth-wiki-tables13,09481.98%90.68%
nym-en3,58482.06%85.38%2.47%
kiji-en7,62384.77%85.39%
synth-secrets-en58187.95%97.07%50.54%
russian-pii-66k4,80588.53%91.05%
hivetrace1,66789.74%93.52%1.53%
synth-secrets-ru58191.74%99.31%52.15%
creddata77694.20%97.55%6.62%
arthur-passwords28098.57%98.57%15.27%
secrets-rules74698.79%99.60%54.66%