Arabic name gender
Arabic script resolves through a consonant skeleton index rather than a guessed transliteration. That design choice is the reason three specific names stopped returning confident wrong answers.
100 free credits every day. No card.
Measured 28 August 2026 on a fixed, reproducible fixture. Not a population sample — a fixed input anyone can rerun against us or against a competitor.
Why the skeleton, and not a spelling guess
Arabic script writes consonants and leaves most short vowels implicit, so a single written form corresponds to a family of Latin spellings. There are two ways to bridge that. You can guess a specific Latin spelling and look it up, or you can look up the consonant skeleton and use the combined evidence of every name that matches it.
The skeleton is the stronger signal, and we have the failures to prove it. When speculative full spellings were given priority, علي resolved through an invented "lya" and came back female; خالدة came back male; نورا came back male. In all three the skeleton was right and the single guessed spelling was wrong.
A single spelling guess is not stronger evidence than the whole set of names it might correspond to. Today a full match is only preferred when it agrees with the skeleton and brings a real sample with it.
Real output, country hint SA
Arabic-script inputs report source: script, which tells you the answer came through the bridge rather than a direct match. That distinction is worth keeping in your pipeline, because the bridge is one more place an error can enter.
| Input | Gender | Probability |
|---|---|---|
| محمد (Mohammed) | male | 100 |
| فاطمة (Fatima) | female | 100 |
| علي (Ali) | male | 98 |
| خالدة (Khalida) | female | 100 |
| نورا (Noura) | female | 100 |
| Mohammed | male | 95 |
Transliteration variance in Latin input
Mohammed, Muhammad, Mohamed, Mohammad and Muhammed are the same name in five documents. Latin-script Arabic names therefore miss more often through spelling variance than through genuine absence. Near-spellings resolve via fuzzy matching and matched_as reports which entry was used.
Where you hold the Arabic script form, send that instead — it collapses the whole variant family into one lookup.
Honest limits
Arabic is our weakest measured coverage among the scripts we do support, and the misses are a data gap rather than a bridge failure: when we answer, we are usually right. Definite articles and compound forms (Abd al-, Abu) attach to the given name in ways that vary by country, and RTL text pasted from documents often carries directional marks that should be stripped before sending.
Questions
Do diacritics matter?
No. Arabic diacritics are preserved rather than stripped globally, and the skeleton index does the matching.
Which country code should I pass?
The one your record actually carries. Naming conventions differ substantially across the Arabic-speaking world, and none of those countries publish registration counts.
Why does one name show observations and another none?
Because a skeleton can land on entries that came from a counted source. Where it does, total_names is non-zero and confidence rises accordingly.
Is Persian or Urdu supported?
They share the script but not the name stock, and we do not measure them separately. Test on your own data before relying on it.
Related pages
Predict gender from a Turkish name. Dotted and dotless i handling, the genuinely unisex names, and why correct Turkish answers are still marked unverified.
Predict gender from an Indian name in romanised or Devanagari form. This is our strongest non-Latin market, and this page shows the measured numbers.
Every name lookup returns a probability, a sample size and a confidence tier. What each one measures, how the tiers are set, and where to put your threshold.
A reproducible benchmark across 4,610 names, 25 countries and nine writing systems. Two scripts return almost nothing, and we publish those numbers too.
Reading the gender field alone discards everything that says whether to believe it. What each response field means and the thresholds behind confidence.
Check it against your own list
Every number on this page is reproducible with a free key. If your data breaks it, that is the more interesting result.