Eight names in one pass
Zoroastrian names come mostly from two communities that split apart centuries ago and rarely borrowed from each other. Parsi families in western India kept Old Iranian forms that had passed through Gujarati mouths and Gujarati spelling. Iranian Zoroastrians in Yazd and Kerman kept a smaller stock, and many changed it only in the twentieth century.
Eight names per run, five runs on this page, 40 names in total. A first run takes a few seconds, and the same sentence returns almost instantly the second time. Of those, thirty-two are one word, one is two words and seven are three words or longer.
Describe who the name is for
Optional. A place, a decade, a family, or the register you’re writing in. One specific detail steers the list better than a paragraph of backstory.
Pick a filter if you have one. It narrows the register, not the meaning.
Names, one per line. Nothing is stored.
Names are generated on the server and shown once. Reloading clears them.
A sample of forty
Nothing below has been edited except the trailing explanation. Most of these lines came back as a name, a dash and a sentence describing it, and the sentence is cut here so the set can be read as a list.
Small town: Gulnaz, Parvin, Shirin, Mehrbanu, Yasmin, Roxana, Farangis, Dilaram.
Coastal family: Dastur Hormazdyar Meherjirana, Rustomji Pestonji Wadia, Ardeshir Behramji Malabari, Framji Cowasji Banaji, Sorabji Shapurji Bengali, Khurshedji Nusserwanji Cama, Dadabhai Naoroji, Jamsetji Nasarwanji Tata.
Modern city: Farrokh, Yasmin, Rustom, Parisa, Shirin, Kaveh, Anahita, Zubin.
Grandparent: Dinyar, Meherban, Rustom, Gulnar, Behram, Parvin, Faridoon, Avabai.
Two languages: Farzaneh, Dariush, Shirin, Kaveh, Parisa, Rostam, Sepand, Golnaz.
The shape of forty names
Of the forty names, thirty-two are one word, one are two words and seven run to three or more. Thirty-two of them arrive with a description attached.
No closing word repeats more than three times, which is why the set reads as forty separate names rather than one name with variations.
Parsi surnames like Tata, Wadia and Mistry came from trades, and the given names alongside them came from scripture. A priest's son was often named for a yazata, one of the beings honoured in prayer, and the name carried no claim beyond that. Gujarati speakers wrote Avestan sounds with Gujarati letters, and the English alphabet then mangled the result again. Ratanbai, Dosabhai and similar forms show the Gujarati habit of attaching -bai or -bhai to a stock name. Iranian Zoroastrians used names like Goshtasb, Esfandiar and Sohrab that also appear in the Shahnameh. That overlap means a name can be Zoroastrian and Persian at once, with no way to tell from the word alone. Families who moved to Tehran in the 1930s sometimes shortened a name to its first two syllables for daily use. Migration to Britain, Canada and the Gulf added another spelling layer on top of the Indian and Iranian ones. A name registered in Toronto or London often keeps the Indian spelling while the same family in Mumbai uses another. For a wider look at how naming traditions get rebuilt from fragments, the Celtic God Name Generator covers a similar problem.
Scriptural and priestly — names lifted from prayer and given to sons of clergy. Avestan forms. Parsi mercantile — trade names and Gujarati endings, common in Bombay and Surat families. Dosabhai. Iranian classical — names shared with the Shahnameh and still current in Yazd and Tehran. Esfandiar. Compound devotional — two good words joined, the oldest layer still in daily use. Meherban. Modern shortened — full names cut to two syllables for school and work. Firoz. Diaspora respelled — the same name written for a passport office abroad. Rustomji. For contrast, the Fate/Grand Order Name Generator shows what happens when a naming pool is invented rather than inherited.
Failures, and the prompt behind each
a zoroastrian name that is very old and very tired returned Zarathushtra, Pouruchista, Hutaosa, Jamaspa, Frasastri, Vidrangha, Daeva, Arshtat. No ending repeats across the eight. None of the eight is wrong, and none of them is about your subject in particular. That is the failure mode worth recognising.
a zoroastrian name for something that guards a door returned Darbān, Pāsbān, Darōbān, Pāsdār, Bānūg, Aiwisrūthrem, Srōsh, Ratu. None of the eight ends the way another one does. That is eight names for the object, not for the subject. The sentence handed the model a noun and the noun ate the set.
Read the two prompts again and the two failures are exact opposites. One gives the model a tone to imitate and nothing to name. The other gives it a single object to name and no tone at all. Both return eight plausible names, and neither returns eight names for your subject in particular.
The three ways, measured
Three routes get a zoroastrian name. Only one of them is a name generator.
| What matters | By hand | A general chatbot | This generator |
|---|---|---|---|
| Time for eight names | Twenty minutes or more | One prompt, then editing | 652 ms per set, instant on a repeat |
| Same prompt, same eight | Only if you write it down | No, the wording drifts | Yes |
| Explains the choice | You do | Yes, at length | No |
| Cost | Your time | A subscription | Free, no account |
| Checks trademarks | No | No | No |
A first run takes a few seconds. The same sentence again returns in milliseconds, because the backend caches each set by prompt, so a set you liked is reproducible tomorrow and a set you did not like costs nothing to replace.
It can't tell you whether a name is currently fashionable among Parsi families in Mumbai or considered old-fashioned. It won't confirm that a name belongs to a specific family, since surnames and given names travel separately. Anything you type about meaning gets ignored, because the literal sense of most of these names is not what people hear.
How to fix a bad set
a zoroastrian name for a girl born in a small town in the 1950s is the prompt that produced the set above. It works because it commits to something you could point at, and the eight that came back are Gulnaz, Parvin, Shirin, Mehrbanu, Yasmin.
You type something like "Parsi girl, born in Bombay, family from Navsari, want something old but usable today. It reads for community words, gender, era and length preference, and it drops anything about personality or mood. A description heavy on feeling and light on place tends to produce names that could belong to any Iranian family. Mention a city and the output shifts, because Bombay, Yazd and Toronto carry different naming habits. Say "short" and you get two-syllable names; say "traditional" and the compounds come back. It ignores requests for a meaning, since most of these names lost their literal sense in daily use long ago. You get eight names, not ranked, and no explanation of why each one appeared. Run the same sentence twice and the list changes, which is worth knowing before you get attached to one. Names it can't place confidently get left out rather than guessed at, so the list stays short of oddities. Nothing in the output tells you whether a name is Parsi or Iranian, and often it belongs to both.
What people want to know
Are these names only used by Parsis?
No, many of them are shared with Iranian Muslims and with Persian speakers generally. Goshtasb and Esfandiar appear in the Shahnameh, which is read far beyond Zoroastrian families. A name alone won't tell you someone's religion, and it never did.
Can I use a name from the list for a child?
Yes, and parents do, though you should check the spelling with family first. The list gives you candidates, not a decision, and it doesn't know your family's conventions. Registration is a separate step with its own rules.
Why did I get eight names and not more?
Eight is the fixed output, so the list stays readable and comparable between runs. The backend caches each set by prompt, so the same sentence returns the same eight names; change one clause and the set changes. If none fit, change the description rather than the tool.
Do the names have meanings?
Most had a literal sense once, built from parts like meher or yar. In daily use that sense has faded, and people hear a name rather than a phrase. The output won't give you a meaning, because the honest answer is often unclear.
Which spelling should I use?
Use the one already on your family's documents if there is one. Otherwise pick a spelling that reads the same way to strangers and to relatives. Consistency across documents matters more than matching anyone else's version.
