Maker

The technical map of ElevenLabs audio infrastructure

ElevenLabs is the best known AI voice infrastructure and runs more than ten operational units: text to speech, speech to text, voice conversion, sound effects and music. The parameter that matters most for a Persian-speaking reader is this: Persian processing is logged only on the language list of the v3 unit.

  • Persian on v3 only
  • Free plan issues no commercial licence
  • Terms of use name Iran

Last checked: This is a reference page. It is re-checked against the vendor sources and updated when a new version ships. Change log

Products

The unit a user logs into, as distinct from the model identifiers called in the API.

Versions

Versions Deployed Status Model families
ElevenLabs Flash v2.5 current
ElevenLabs Multilingual v2 current
ElevenLabs v3 current

Which unit supports Persian processing

ElevenLabs runs three primary text-to-speech units and their language support lists are not identical. v3 carries a 74-language list with Persian logged as fas. Flash v2.5 carries 32 languages and Multilingual v2 carries 29, and Persian is logged on neither.

The critical point: on operating parameters, Flash v2.5 looks like the more attractive unit — 75ms latency, a 40,000-character ceiling against v3 5,000, and a lower rate. A user who selects this unit and feeds it Persian text receives a degraded output and reads it as an infrastructure fault. The actual cause is unit misselection, not a system limitation.

The free plan does not grant what it appears to

The free plan issues 10,000 operational credits a month, sufficient for evaluation. But the same pricing page states that this plan carries neither a commercial licence nor voice cloning.

So a whole project can be tested on the free plan, the audio output extracted, and only then does it emerge that no licence was issued to use that output in revenue-generating work. Instant cloning is issued starting on the Starter tier at $6 a month, professional cloning on the Creator tier at $22, and the commercial licence is issued from those same tiers.

The infrastructure is not limited to voice generation

Scribe v2 is this infrastructure speech-to-text unit, and its language support list exceeds ninety languages — several multiples of the same company voice-generation units. Its realtime version carries roughly 150ms of latency. Beyond that, a voice-conversion unit, a sound-effects unit and a music-generation unit also run on this infrastructure.

So if the task is transcribing a Persian interview rather than generating speech, the correct route is Scribe, not the units in this table. That is a distinct operation with a distinct language support list.

The terms of use document Iran by name

The export control clause in the ElevenLabs terms of use requires the user to warrant that they are not located in a country under US economic sanctions, and Iran is logged in that list. This is read directly from that document.

This report issues no accounts, refers no intermediaries, and describes no route around that clause. If the question is whether an official route exists for use from Iran, the documented answer on this page is that no such route is logged for this infrastructure.

Documented strengths

  • The only infrastructure in this category that ships new voice generation from a sample together with a commercial licence
  • v3 carries a 74-language list with Persian logged on it
  • Flash v2.5, at 75ms latency and a 40,000-character ceiling, is genuinely fast for high-volume English-language operational load
  • The transcription unit carries a language support list exceeding ninety, well beyond the same company voice-generation units

Known weaknesses and limits

  • Persian processing is logged on the v3 list only. The other two text-to-speech units do not support it, and one of them appears the more attractive default.
  • The free plan issues neither a commercial licence nor voice cloning, so free evaluation output cannot be deployed in revenue-generating work.
  • The terms of use document Iran by name in the export control clause.
  • No Persian-specific voice is documented, meaning Persian text is processed through a voice not engineered for that language.

Technical verdict

For Persian voice production, this is the only unit that carries both language support and a custom-voice licence, provided the unit is manually set to v3 and it is understood that the free plan issues no commercial rights.

Frequently raised questions, with documented answers

Does ElevenLabs support Persian

Only on the v3 unit. Its 74-language list logs Persian as fas. Flash v2.5 and Multilingual v2 carry 32 and 29 languages respectively, and Persian is on neither.

Can the free plan be used commercially

No. The pricing page states the free plan carries no commercial licence and no voice cloning. The commercial licence is issued starting on the Starter tier at $6 a month.

Which tier activates voice cloning

Instant cloning from the Starter tier ($6) and professional cloning from the Creator tier ($22). Higher tiers grant more professional-clone capacity.

Sources

  1. ElevenLabs, modelsvendor sourceelevenlabs.ioread on 28 August 2026
  2. ElevenLabs, text to speech languagesvendor sourceelevenlabs.ioread on 28 August 2026
  3. ElevenLabs, pricingvendor sourceelevenlabs.ioread on 28 August 2026
  4. ElevenLabs, terms of usevendor sourceelevenlabs.ioread on 28 August 2026