Orpheus Arabic Saudi (on Groq)

Orpheus Arabic Saudi is a text-to-speech model from Canopy Labs served through Groq's inference API, and the only model we found whose name itself says Saudi. It replaced Groq's previous Arabic voice offering when the platform switched its speech models on 30 January 2026, and two more voices were added on 18 April 2026. The model exposes six Saudi-dialect voices, Abdullah, Fahad, Sultan, Lulwa, Noura and Aisha, through the same API developers use for Groq's language models, the quickest way to hear a Saudi voice from code without a separate vendor account. The constraints are real: the model is in preview, each request is limited to two hundred characters, and Groq documents nothing about diacritics, number reading, mixed Arabic-English text or voice cloning. Processing happens in Groq's data centres, none of which it lists in the Gulf. No independent listening score exists for it. Pricing in November 2026: $40 per million characters, billed through a Groq developer account.