FastPII Docs
Benchmarks

Performance Benchmarks

Latency, throughput, memory, and false-positive characteristics for FastPII.

Performance Benchmarks

FastPII is designed for production pipelines where privacy checks must be fast enough to sit inline with APIs, batch jobs, and AI preprocessing.

Headline metrics

  • <15ms per document
  • 7,200 docs/second throughput
  • ~50MB memory footprint
  • <1% false positive rate with checksum validation

What these numbers mean in practice

  • API gateways can scan requests without adding major latency.
  • Batch ingestion jobs can process large corpora quickly.
  • Small memory usage makes local deployment and serverless packaging easier.
  • Checksum validation helps keep review queues focused on real findings.

Suggested production usage

FastPII is a strong fit for:

  • Pre-prompt sanitization
  • RAG document ingestion
  • CSV and text export cleanup
  • Inline FastAPI request handling

For the best outcome, validate your own workload with representative Czech documents and track latency separately for detection and transformation stages.

Benchmark Methodology

How FastPII detector accuracy and false-positive rates were evaluated.

GDPR Article 9 Detection

Handle rodné číslo findings that may reveal special category data under GDPR Article 9.

On this page

Performance BenchmarksHeadline metricsWhat these numbers mean in practiceSuggested production usage