278M PII detection model bests OpenAI, AWS and Microsoft
Hacker News · 4 Nov 2026, 05:02 AEST
Protegrity-PII is a new PII detection model that outperforms OpenAI, AWS, and Microsoft. It achieves a typed recall of 0.910 and a precision of 0.902 on the PII-SCOPE benchmark.
Key facts
- Protegrity-PII achieves a typed recall of 0.910 with a precision of 0.902 on a new benchmark we propose, PII-SCOPE.
- Our model processes text in 0.49 seconds per million characters at $0.0005 per million characters on a single RTX 6000 using batched inference.
- Due in part to the expansive nature of the entity type taxonomy used in training, Protegrity-PII achieves superior out-of-distribution generalization when benchmarked against a selection of state of the art models and datasets.
Read aloud
Protegrity-PII is a new PII detection model that outperforms Open A I, AWS, and Microsoft. It achieves a typed recall of zero point nine one zero and a precision of zero point nine zero two on the PII-SCOPE benchmark. Protegrity-PII achieves a typed recall of zero point nine one zero with a precision of zero point nine zero two on a new benchmark we propose, PII-SCOPE. Our model processes text in zero point four nine seconds per million characters at zero point zero zero zero five dollars per million characters on a single RTX six thousand using batched inference. Due in part to the expansive nature of the entity type taxonomy used in training, Protegrity-PII achieves superior out-of-distribution generalization when benchmarked against a selection of state of the art models and datasets.
Sources in this cluster
- 278M PII detection model bests OpenAI, AWS and Microsoft
Hacker News · 4 Nov 2026, 05:02 AEST