gpt-4o-mini: 1 credit — fast, handles most documents well. gpt-4o: 2 credits — higher accuracy for complex or multi-person cases.
Advanced settings
Mode
Standard: regex patterns + LLM scan for names/orgs/places. Strict: also replaces any capitalised two-word phrase as a potential name — more aggressive, may produce false positives.
Region
Nordic: Norwegian fødselsnummer, phone, email, addresses. European: adds IBAN, SE personnummer, UK NI. ECHR: adds application numbers, DOB phrases. Global: adds US SSN, document numbers.
Redact
Officials
When checked, judges, expert witnesses and caseworkers keep their names in a labelled tag: [JUDGE: Andersen]. Uncheck to replace all names with generic role tags.
Output
Contextual: each person gets a role tag so their identity is traceable within the document. Generic: all names become [PERSON]. Pseudonyms: replaced with plausible fake Norwegian values.
Names listed here will never be redacted, even if the AI would otherwise remove them — e.g. a judge or expert who must remain identifiable.
Replace a specific name with a custom bracketed label, e.g. “David Jr” → [Junior].