KlusAI KlusAI Research

Flagship benchmark

EuroPriv-Bench Leaderboard

Detection ≠ protection: compare leakage on each config, not just detection F1.

For privacy, the number that matters is re-identification leakage — how many decode-bearing national IDs a model leaves un-redacted — not detection F1. A leaked ID silently discloses identifying attributes (a Romanian CNP encodes date of birth, sex and county), so the best detector is not necessarily the best protector. Config status is shown on each row: development results and externally annotated gold have different evidence boundaries. Programme panel validation is in progress; no existing row is promoted by this page.

October update: klusai/kp-deid-xlmr-560m-legal is a legal specialist. Its TAB DIRECT result does not establish general privacy protection: on ro-synthetic-v1 it misses every CNP. Both results, with confidence intervals and subject counts, are below. See the October update for the comparison with stock Presidio, seed counts and CPU timing limits.

Scatter plot of detection F1 (x-axis) against CNP re-identification leak rate (y-axis) for eight models on ro-realskeleton-v1; higher F1 does not imply lower leakage.
Detection–protection dissociation: detection F1 (x) vs CNP re-identification leak rate (y) on ro-realskeleton-v1, dev status, n=1123 CNPs. Historical figure; the newly synced 560m legal rows appear in the tables below. Higher F1 does not imply lower leakage.

TAB direct-identifier leakage — legal specialist

TAB is external English legal gold. This table renders the uniform board’s tab_reid_leakage rows; models without that metric in the synced data are omitted. The historical metric checks whether any token is redacted in every occurrence of a direct identifier. It can count partial masking as success, so it is a proxy rather than a complete residual-removal guarantee or measured attacker re-identification probability. Rates and Wilson confidence intervals are percentages; n counts DIRECT subjects, not documents. The legal row is the seed0 checkpoint; its CI does not measure training-seed uncertainty.

Model Track Contamination Config status DIRECT leak rate 95% CI Subjects leaked / n Documents
klusai/kp-deid-xlmr-560m-legal tab-echr-legal-en-v1 clean_held_out real-external-gold 9.47% 6.5–13.61% 25 / 264 127
klusai/kos-shield@763956e6da84eb755c168acf13c5d10b9239ca5c5415a8a30b4c806c8c55dd1f tab-echr-legal-en-v1 clean_held_out real-external-gold 50.38% 44.39–56.36% 133 / 264 127

National-ID leakage — by config

Detection F1 is not privacy. EuroPriv-Bench measures re-identification leakage: a missed (un-redacted) national ID deterministically discloses identifying attributes — on the Romanian configs a leaked CNP discloses date of birth + sex + county, on the Polish track a leaked PESEL discloses date of birth + sex, and on the Italian track a leaked Codice Fiscale discloses date of birth + sex + place of birth. The bar shows the leak rate (long/red = leaks, tiny/green ≈ protects); the table also counts national IDs left un-redacted and the quasi-identifiers thereby leaked (lower is better).

Model Track Contamination Config status Leak rate 95% CI IDs missed / n Quasi-identifiers
fastino/gliner2-base-v1 legal-realskeleton-v1 clean held-out dev 65.4% 62.96–67.77% 981 / 1500 2943
fastino/gliner2-base-v1 ro-realskeleton-v1 clean held-out dev 28.6% 26.02–31.3% 321 / 1123 963
fastino/gliner2-base-v1 ro-synthetic-v1 clean held-out dev 8.1% 6.54–9.9% 82 / 1017 246
fastino/gliner2-base-v1 it-realskeleton-v1 clean held-out dev 39.2% 36.33–42.06% 435 / 1111 1305
fastino/gliner2-base-v1 pl-realskeleton-v1 clean held-out dev 33.8% 31.02–36.61% 370 / 1096 740
urchade/gliner_multi_pii-v1 legal-realskeleton-v1 clean held-out dev 71.7% 69.4–73.95% 1076 / 1500 3228
urchade/gliner_multi_pii-v1 ro-realskeleton-v1 clean held-out dev 30.2% 27.57–32.94% 339 / 1123 1017
urchade/gliner_multi_pii-v1 ro-synthetic-v1 unknown dev 0.0% 0.0–0.38% 0 / 1017 0
urchade/gliner_multi_pii-v1 it-realskeleton-v1 clean held-out dev 36.2% 33.41–39.05% 402 / 1111 1206
urchade/gliner_multi_pii-v1 pl-realskeleton-v1 clean held-out dev 57.8% 54.9–60.74% 634 / 1096 1268
klusai/kp-deid-mdeberta-280m legal-realskeleton-v1 clean held-out dev 4.1% 3.18–5.19% 61 / 1500 183
klusai/kp-deid-mdeberta-280m ro-realskeleton-v1 clean held-out dev 0.0% 0.0–0.34% 0 / 1123 0
klusai/kp-deid-mdeberta-280m ro-synthetic-v1 in-distribution dev 0.0% 0.0–0.38% 0 / 1017 0
klusai/kp-deid-mdeberta-280m it-realskeleton-v1 clean held-out dev 0.0% 0.0–0.34% 0 / 1111 0
klusai/kp-deid-mdeberta-280m pl-realskeleton-v1 clean held-out dev 0.0% 0.0–0.35% 0 / 1096 0
klusai/kp-deid-xlmr-560m legal-realskeleton-v1 clean held-out dev 0.0% 0.0–0.26% 0 / 1500 0
klusai/kp-deid-xlmr-560m ro-realskeleton-v1 clean held-out dev 0.0% 0.0–0.34% 0 / 1123 0
klusai/kp-deid-xlmr-560m ro-synthetic-v1 in-distribution dev 0.0% 0.0–0.38% 0 / 1017 0
klusai/kp-deid-xlmr-560m pl-realskeleton-v1 clean held-out dev 0.0% 0.0–0.35% 0 / 1096 0
klusai/kp-deid-xlmr-560m-legal ro-realskeleton-v1 clean held-out dev 0.0% 0.0–0.34% 0 / 1123 0
klusai/kp-deid-xlmr-560m-legal ro-synthetic-v1 in-distribution dev 100.0% 99.62–100.0% 1017 / 1017 3051
klusai/kp-deid-xlmr-560m-legal pl-realskeleton-v1 clean held-out dev 0.0% 0.0–0.35% 0 / 1096 0
OpenMed/privacy-filter-multilingual legal-realskeleton-v1 clean held-out dev 83.9% 81.92–85.64% 1258 / 1500 3774
OpenMed/privacy-filter-multilingual ro-realskeleton-v1 clean held-out dev 26.4% 23.86–29.01% 296 / 1123 888
OpenMed/privacy-filter-multilingual ro-synthetic-v1 unknown dev 1.9% 1.2–2.9% 19 / 1017 57
OpenMed/privacy-filter-multilingual it-realskeleton-v1 clean held-out dev 35.4% 32.62–38.23% 393 / 1111 1179
OpenMed/privacy-filter-multilingual pl-realskeleton-v1 clean held-out dev 3.7% 2.77–5.04% 41 / 1096 82
presidio-analyzer+en_core_web_lg legal-realskeleton-v1 clean held-out dev 66.5% 64.04–68.81% 997 / 1500 2991
presidio-analyzer+en_core_web_lg ro-realskeleton-v1 clean held-out dev 0.0% 0.0–0.34% 0 / 1123 0
presidio-analyzer+en_core_web_lg ro-synthetic-v1 clean held-out dev 0.0% 0.0–0.38% 0 / 1017 0
presidio-analyzer+en_core_web_lg it-realskeleton-v1 clean held-out dev 96.5% 95.24–97.42% 1072 / 1111 3216
presidio-analyzer+en_core_web_lg pl-realskeleton-v1 clean held-out dev 0.0% 0.0–0.35% 0 / 1096 0
openai/privacy-filter legal-realskeleton-v1 clean held-out dev 96.7% 95.63–97.46% 1450 / 1500 4350
openai/privacy-filter ro-realskeleton-v1 clean held-out dev 1.4% 0.88–2.3% 16 / 1123 48
openai/privacy-filter ro-synthetic-v1 unknown dev 0.1% 0.02–0.55% 1 / 1017 3
openai/privacy-filter it-realskeleton-v1 clean held-out dev 0.4% 0.14–0.92% 4 / 1111 12
openai/privacy-filter pl-realskeleton-v1 clean held-out dev 0.1% 0.02–0.52% 1 / 1096 2
spacy/en_core_web_lg@3.8.0 legal-realskeleton-v1 clean held-out dev 66.0% 63.56–68.35% 990 / 1500 2970
spacy/en_core_web_lg@3.8.0 ro-realskeleton-v1 clean held-out dev 89.0% 87.09–90.74% 1000 / 1123 3000
spacy/en_core_web_lg@3.8.0 ro-synthetic-v1 clean held-out dev 91.0% 89.03–92.57% 925 / 1017 2775
spacy/en_core_web_lg@3.8.0 it-realskeleton-v1 clean held-out dev 39.2% 36.33–42.06% 435 / 1111 1305
spacy/en_core_web_lg@3.8.0 pl-realskeleton-v1 clean held-out dev 44.7% 41.79–47.66% 490 / 1096 980
tabularisai/eu-pii-safeguard legal-realskeleton-v1 clean held-out dev 27.3% 25.07–29.58% 409 / 1500 1227
tabularisai/eu-pii-safeguard ro-realskeleton-v1 clean held-out dev 35.4% 32.61–38.19% 397 / 1123 1191
tabularisai/eu-pii-safeguard ro-synthetic-v1 unknown dev 0.0% 0.0–0.38% 0 / 1017 0
tabularisai/eu-pii-safeguard it-realskeleton-v1 clean held-out dev 34.4% 31.65–37.23% 382 / 1111 1146
tabularisai/eu-pii-safeguard pl-realskeleton-v1 clean held-out dev 30.7% 28.0–33.45% 336 / 1096 672
klusai/kos-shield@763956e6da84eb755c168acf13c5d10b9239ca5c5415a8a30b4c806c8c55dd1f ro-realskeleton-v1 clean held-out dev 100.0% 99.66–100.0% 1123 / 1123 3369
klusai/kos-shield@763956e6da84eb755c168acf13c5d10b9239ca5c5415a8a30b4c806c8c55dd1f pl-realskeleton-v1 clean held-out dev 100.0% 99.65–100.0% 1096 / 1096 2192

National-ID leak rates and Wilson confidence intervals are percentages; n counts identifier subjects. Development configs remain dev while panel validation is in progress. A zero point estimate is not proof of zero risk. Compare models on the same held-out config; a specialist's result does not establish performance across languages or domains.

Detection scores — by model and language

Entity-level scores on the klusai/europriv-bench test split, by model and language. Higher F1 is better; the table defaults to best-first. Click a column header to re-sort. Rows where the model was trained on the config’s own source data are greyed (in-distribution) — their scores are inflated by train/eval overlap and are not a fair test.

How to read this — contamination & validation

Each row carries two governance markers. Contamination flags whether the model was trained on that config's source data — an in-distribution score is inflated by train/eval overlap (e.g. a perfect 100/100/100 is a memorisation artefact, not a win), while a clean held-out score is a fair test. Config status preserves the source's config_status value. dev means validation is incomplete; real-external-gold identifies externally annotated real data. Future panel-validated rows must pass the pre-registered LLM-panel protocol; that label does not imply native-speaker review. Legacy citable-validated rows retain their source status and protocol. Unknown values remain visible, without inferred validation.

Each row reports entity-level precision / recall / F1 (×100) under the unified KlusAI privacy taxonomy. Results carry full provenance (model id, dataset config/split, harness & taxonomy version, timestamp) in the source repository.

Schema v3 · Benchmark v0.2.0 · Taxonomy v0.2.0

Model Adapter Track Lang Domain Precision Recall F1 n Contamination Config status
fastino/gliner2-base-v1 gliner2 legal-realskeleton-v1 ro legal 82.5 55.9 66.7 1500 clean held-out dev
fastino/gliner2-base-v1 gliner2 ro-realskeleton-v1 ro legal 74.6 56.4 64.2 1500 clean held-out dev
fastino/gliner2-base-v1 gliner2 ro-synthetic-v1 ro general 80.2 77.3 78.7 1500 clean held-out dev
fastino/gliner2-base-v1 gliner2 it-realskeleton-v1 it legal 87.3 69.2 77.2 1500 clean held-out dev
fastino/gliner2-base-v1 gliner2 pl-realskeleton-v1 pl legal 57.4 47.4 51.9 1500 clean held-out dev
fastino/gliner2-base-v1 gliner2 nl nl general 50.7 42.8 46.4 1500 clean held-out dev
fastino/gliner2-base-v1 gliner2 en en general 58.4 45.3 51.0 1500 clean held-out dev
fastino/gliner2-base-v1 gliner2 fr fr general 56.6 47.6 51.7 1500 clean held-out dev
fastino/gliner2-base-v1 gliner2 de de general 56.1 47.1 51.2 1500 clean held-out dev
fastino/gliner2-base-v1 gliner2 it it general 54.6 42.8 48.0 1500 clean held-out dev
fastino/gliner2-base-v1 gliner2 es es general 56.8 47.6 51.8 1500 clean held-out dev
urchade/gliner_multi_pii-v1 gliner legal-realskeleton-v1 ro legal 89.3 72.9 80.3 1500 clean held-out dev
urchade/gliner_multi_pii-v1 gliner ro-realskeleton-v1 ro legal 93.9 78.2 85.3 1500 clean held-out dev
urchade/gliner_multi_pii-v1 gliner ro-synthetic-v1 ro general 82.2 80.4 81.3 1500 unknown dev
urchade/gliner_multi_pii-v1 gliner it-realskeleton-v1 it legal 92.1 80.0 85.6 1500 clean held-out dev
urchade/gliner_multi_pii-v1 gliner pl-realskeleton-v1 pl legal 93.6 73.7 82.5 1500 clean held-out dev
urchade/gliner_multi_pii-v1 gliner nl nl general 63.9 50.9 56.7 1500 unknown dev
urchade/gliner_multi_pii-v1 gliner en en general 61.9 42.0 50.0 1500 unknown dev
urchade/gliner_multi_pii-v1 gliner fr fr general 65.6 48.7 55.9 1500 unknown dev
urchade/gliner_multi_pii-v1 gliner de de general 64.1 51.7 57.2 1500 unknown dev
urchade/gliner_multi_pii-v1 gliner it it general 64.5 46.6 54.1 1500 unknown dev
urchade/gliner_multi_pii-v1 gliner es es general 66.2 47.4 55.2 1500 unknown dev
klusai/kp-deid-mdeberta-280m kp-model legal-realskeleton-v1 ro legal 51.5 71.8 60.0 1500 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model ro-realskeleton-v1 ro legal 68.6 80.5 74.1 1500 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model ro-synthetic-v1 ro general 100.0 100.0 100.0memorised 1500 in-distribution dev
klusai/kp-deid-mdeberta-280m kp-model it-realskeleton-v1 it legal 62.1 79.7 69.8 1500 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model pl-realskeleton-v1 pl legal 71.0 82.3 76.3 1500 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model default cs general 44.4 41.3 42.8 399 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model default nl general 35.7 33.4 34.5 399 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model nl nl general 57.0 48.4 52.3 1500 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model en en general 51.8 41.9 46.4 1500 unknown dev
klusai/kp-deid-mdeberta-280m kp-model default fi general 38.5 35.1 36.7 399 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model default fr general 35.6 34.2 34.9 398 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model fr fr general 56.0 46.7 50.9 1500 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model default de general 31.0 29.2 30.1 399 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model de de general 56.1 46.2 50.7 1500 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model default el general 35.5 33.5 34.4 397 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model default hu general 30.2 29.3 29.7 400 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model it it general 50.6 41.5 45.6 1500 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model default pl general 35.2 32.9 34.0 397 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model default ro general 35.5 33.7 34.6 400 unknown dev
klusai/kp-deid-mdeberta-280m kp-model es es general 51.2 42.6 46.5 1500 clean held-out dev
klusai/kp-deid-mdeberta-280m kp-model default sv general 38.2 36.3 37.2 397 clean held-out dev
klusai/kp-deid-xlmr-560m kp-model legal-realskeleton-v1 ro legal 50.7 62.4 55.9 1500 clean held-out dev
klusai/kp-deid-xlmr-560m kp-model ro-realskeleton-v1 ro legal 63.1 71.9 67.2 1500 clean held-out dev
klusai/kp-deid-xlmr-560m kp-model ro-synthetic-v1 ro general 100.0 100.0 100.0memorised 1500 in-distribution dev
klusai/kp-deid-xlmr-560m kp-model pl-realskeleton-v1 pl legal 69.4 78.6 73.7 1500 clean held-out dev
klusai/kp-deid-xlmr-560m kp-model nl nl general 58.4 53.3 55.7 1500 clean held-out dev
klusai/kp-deid-xlmr-560m kp-model en en general 52.7 45.9 49.1 1500 unknown dev
klusai/kp-deid-xlmr-560m kp-model fr fr general 57.0 51.7 54.2 1500 clean held-out dev
klusai/kp-deid-xlmr-560m kp-model de de general 57.6 53.0 55.2 1500 clean held-out dev
klusai/kp-deid-xlmr-560m kp-model it it general 54.8 47.6 50.9 1500 clean held-out dev
klusai/kp-deid-xlmr-560m kp-model es es general 55.9 50.6 53.1 1500 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model ro-realskeleton-v1 ro legal 44.8 51.9 48.1 1500 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model ro-synthetic-v1 ro general 54.6 42.6 47.9 1500 in-distribution dev
klusai/kp-deid-xlmr-560m-legal kp-model pl-realskeleton-v1 pl legal 44.9 51.7 48.1 1500 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model default cs general 47.6 27.0 34.5 399 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model default nl general 43.5 24.5 31.3 399 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model nl nl general 60.1 36.5 45.4 1500 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model en en general 55.8 31.7 40.4 1500 unknown dev
klusai/kp-deid-xlmr-560m-legal kp-model default fi general 42.7 21.3 28.4 399 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model default fr general 45.4 26.4 33.4 398 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model fr fr general 56.9 34.9 43.2 1500 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model default de general 43.3 24.1 30.9 399 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model de de general 61.0 35.8 45.2 1500 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model default el general 40.6 22.7 29.1 397 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model default hu general 42.5 23.8 30.5 400 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model it it general 55.8 32.3 40.9 1500 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model default pl general 43.5 24.5 31.3 397 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model default ro general 44.9 25.3 32.4 400 unknown dev
klusai/kp-deid-xlmr-560m-legal kp-model es es general 50.6 27.9 36.0 1500 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model default sv general 42.2 23.2 29.9 397 clean held-out dev
klusai/kp-deid-xlmr-560m-legal kp-model tab-echr-legal-en-v1 en legal 39.2 30.0 34.0 127 clean held-out real-external-gold
OpenMed/privacy-filter-multilingual openmed legal-realskeleton-v1 ro legal 55.7 44.3 49.3 1500 clean held-out dev
OpenMed/privacy-filter-multilingual openmed ro-realskeleton-v1 ro legal 63.8 52.5 57.6 1500 clean held-out dev
OpenMed/privacy-filter-multilingual openmed ro-synthetic-v1 ro general 71.8 76.4 74.1 1500 unknown dev
OpenMed/privacy-filter-multilingual openmed it-realskeleton-v1 it legal 66.5 64.7 65.6 1500 clean held-out dev
OpenMed/privacy-filter-multilingual openmed pl-realskeleton-v1 pl legal 73.1 60.5 66.2 1500 clean held-out dev
OpenMed/privacy-filter-multilingual openmed nl nl general 69.8 57.6 63.1 1500 in-distribution dev
OpenMed/privacy-filter-multilingual openmed en en general 66.3 54.7 59.9 1500 in-distribution dev
OpenMed/privacy-filter-multilingual openmed fr fr general 66.7 56.3 61.1 1500 in-distribution dev
OpenMed/privacy-filter-multilingual openmed de de general 67.0 55.6 60.8 1500 in-distribution dev
OpenMed/privacy-filter-multilingual openmed it it general 60.7 50.3 55.0 1500 in-distribution dev
OpenMed/privacy-filter-multilingual openmed es es general 66.0 53.5 59.1 1500 in-distribution dev
presidio-analyzer+en_core_web_lg presidio legal-realskeleton-v1 ro legal 42.9 43.9 43.4 1500 clean held-out dev
presidio-analyzer+en_core_web_lg presidio ro-realskeleton-v1 ro legal 47.2 47.3 47.2 1500 clean held-out dev
presidio-analyzer+en_core_web_lg presidio ro-synthetic-v1 ro general 54.7 55.9 55.3 1500 clean held-out dev
presidio-analyzer+en_core_web_lg presidio it-realskeleton-v1 it legal 51.5 48.0 49.6 1500 clean held-out dev
presidio-analyzer+en_core_web_lg presidio pl-realskeleton-v1 pl legal 45.5 49.3 47.3 1500 clean held-out dev
presidio-analyzer+en_core_web_lg presidio default cs general 27.4 28.4 27.9 399 clean held-out dev
presidio-analyzer+en_core_web_lg presidio default nl general 29.3 29.1 29.2 399 clean held-out dev
presidio-analyzer+en_core_web_lg presidio nl nl general 22.4 23.1 22.7 1500 clean held-out dev
presidio-analyzer+en_core_web_lg presidio en en general 55.8 35.9 43.7 1500 clean held-out dev
presidio-analyzer+en_core_web_lg presidio default fi general 26.3 31.1 28.5 399 clean held-out dev
presidio-analyzer+en_core_web_lg presidio default fr general 42.8 28.7 34.3 398 clean held-out dev
presidio-analyzer+en_core_web_lg presidio fr fr general 35.4 24.2 28.7 1500 clean held-out dev
presidio-analyzer+en_core_web_lg presidio default de general 21.0 28.2 24.1 399 clean held-out dev
presidio-analyzer+en_core_web_lg presidio de de general 19.8 26.4 22.6 1500 clean held-out dev
presidio-analyzer+en_core_web_lg presidio default el general 34.4 33.3 33.8 397 clean held-out dev
presidio-analyzer+en_core_web_lg presidio default hu general 28.5 30.8 29.6 400 clean held-out dev
presidio-analyzer+en_core_web_lg presidio it it general 25.3 22.0 23.6 1500 clean held-out dev
presidio-analyzer+en_core_web_lg presidio default pl general 25.1 27.4 26.2 397 clean held-out dev
presidio-analyzer+en_core_web_lg presidio default ro general 31.9 28.9 30.3 400 clean held-out dev
presidio-analyzer+en_core_web_lg presidio es es general 28.1 20.6 23.8 1500 clean held-out dev
presidio-analyzer+en_core_web_lg presidio default sv general 25.2 27.8 26.5 397 clean held-out dev
openai/privacy-filter privacy-filter legal-realskeleton-v1 ro legal 37.5 26.8 31.3 1500 clean held-out dev
openai/privacy-filter privacy-filter ro-realskeleton-v1 ro legal 38.8 34.2 36.3 1500 clean held-out dev
openai/privacy-filter privacy-filter ro-synthetic-v1 ro general 56.0 59.2 57.6 1500 unknown dev
openai/privacy-filter privacy-filter it-realskeleton-v1 it legal 71.1 65.4 68.1 1500 clean held-out dev
openai/privacy-filter privacy-filter pl-realskeleton-v1 pl legal 42.6 38.8 40.6 1500 clean held-out dev
openai/privacy-filter privacy-filter nl nl general 59.3 39.1 47.1 1500 unknown dev
openai/privacy-filter privacy-filter en en general 58.0 32.3 41.5 1500 unknown dev
openai/privacy-filter privacy-filter fr fr general 59.2 38.2 46.4 1500 unknown dev
openai/privacy-filter privacy-filter de de general 63.3 41.4 50.0 1500 unknown dev
openai/privacy-filter privacy-filter it it general 56.7 37.5 45.1 1500 unknown dev
openai/privacy-filter privacy-filter es es general 59.3 38.3 46.5 1500 unknown dev
spacy/en_core_web_lg@3.8.0 spacy legal-realskeleton-v1 ro legal 10.9 17.9 13.6 1500 clean held-out dev
spacy/en_core_web_lg@3.8.0 spacy ro-realskeleton-v1 ro legal 14.2 14.3 14.3 1500 clean held-out dev
spacy/en_core_web_lg@3.8.0 spacy ro-synthetic-v1 ro general 30.0 13.5 18.6 1500 clean held-out dev
spacy/en_core_web_lg@3.8.0 spacy it-realskeleton-v1 it legal 15.9 14.3 15.1 1500 clean held-out dev
spacy/en_core_web_lg@3.8.0 spacy pl-realskeleton-v1 pl legal 16.7 14.7 15.7 1500 clean held-out dev
spacy/en_core_web_lg@3.8.0 spacy nl nl general 11.5 9.9 10.7 1500 clean held-out dev
spacy/en_core_web_lg@3.8.0 spacy en en general 48.8 23.4 31.7 1500 clean held-out dev
spacy/en_core_web_lg@3.8.0 spacy fr fr general 17.8 8.9 11.9 1500 clean held-out dev
spacy/en_core_web_lg@3.8.0 spacy de de general 9.4 10.7 10.0 1500 clean held-out dev
spacy/en_core_web_lg@3.8.0 spacy it it general 11.8 8.4 9.8 1500 clean held-out dev
spacy/en_core_web_lg@3.8.0 spacy es es general 12.1 6.8 8.7 1500 clean held-out dev
tabularisai/eu-pii-safeguard tabularisai legal-realskeleton-v1 ro legal 77.8 50.3 61.1 1500 clean held-out dev
tabularisai/eu-pii-safeguard tabularisai ro-realskeleton-v1 ro legal 90.1 63.7 74.7 1500 clean held-out dev
tabularisai/eu-pii-safeguard tabularisai ro-synthetic-v1 ro general 89.3 86.0 87.6 1500 unknown dev
tabularisai/eu-pii-safeguard tabularisai it-realskeleton-v1 it legal 89.6 61.1 72.7 1500 clean held-out dev
tabularisai/eu-pii-safeguard tabularisai pl-realskeleton-v1 pl legal 86.7 64.3 73.8 1500 clean held-out dev
tabularisai/eu-pii-safeguard tabularisai nl nl general 71.2 56.4 62.9 1500 in-distribution dev
tabularisai/eu-pii-safeguard tabularisai en en general 64.4 42.9 51.5 1500 in-distribution dev
tabularisai/eu-pii-safeguard tabularisai fr fr general 69.7 51.8 59.4 1500 in-distribution dev
tabularisai/eu-pii-safeguard tabularisai de de general 71.7 56.9 63.4 1500 in-distribution dev
tabularisai/eu-pii-safeguard tabularisai it it general 68.9 50.0 57.9 1500 in-distribution dev
tabularisai/eu-pii-safeguard tabularisai es es general 67.7 51.2 58.3 1500 in-distribution dev
klusai/kos-shield@763956e6da84eb755c168acf13c5d10b9239ca5c5415a8a30b4c806c8c55dd1f shield ro-realskeleton-v1 ro legal 37.3 46.2 41.3 1500 clean held-out dev
klusai/kos-shield@763956e6da84eb755c168acf13c5d10b9239ca5c5415a8a30b4c806c8c55dd1f shield pl-realskeleton-v1 pl legal 36.1 51.3 42.4 1500 clean held-out dev
klusai/kos-shield@763956e6da84eb755c168acf13c5d10b9239ca5c5415a8a30b4c806c8c55dd1f shield nl nl general 63.8 42.2 50.8 1500 unknown dev
klusai/kos-shield@763956e6da84eb755c168acf13c5d10b9239ca5c5415a8a30b4c806c8c55dd1f shield en en general 59.7 23.7 34.0 1500 unknown dev
klusai/kos-shield@763956e6da84eb755c168acf13c5d10b9239ca5c5415a8a30b4c806c8c55dd1f shield fr fr general 43.6 33.8 38.1 1500 unknown dev
klusai/kos-shield@763956e6da84eb755c168acf13c5d10b9239ca5c5415a8a30b4c806c8c55dd1f shield de de general 42.9 38.8 40.8 1500 unknown dev
klusai/kos-shield@763956e6da84eb755c168acf13c5d10b9239ca5c5415a8a30b4c806c8c55dd1f shield it it general 32.9 30.9 31.8 1500 unknown dev
klusai/kos-shield@763956e6da84eb755c168acf13c5d10b9239ca5c5415a8a30b4c806c8c55dd1f shield es es general 37.8 35.6 36.7 1500 unknown dev
klusai/kos-shield@763956e6da84eb755c168acf13c5d10b9239ca5c5415a8a30b4c806c8c55dd1f shield tab-echr-legal-en-v1 en legal 14.7 14.2 14.4 127 clean held-out real-external-gold

Each row reports entity-level precision / recall / F1 (×100) under the unified KlusAI privacy taxonomy. Results carry full provenance (model id, dataset config/split, harness & taxonomy version, timestamp) in the source repository.

How to submit

EuroPriv-Bench is open. Run the harness against your model and open a PR adding your entry to baselines/leaderboard.json — see the benchmark repo for the adapter contract and reproduction steps. Entries without reproducible provenance are not listed.