Qwen3 TTS now available on Together AI
Qwen's open text-to-speech model is now on Together AI, on serverless inference and dedicated endpoints, at $20 per 1M characters.
Models:Qwen/Qwen3-TTS-12Hz-1.7B
What changed
Qwen3 TTS is now available on Together AI, on Together Serverless Inference and on dedicated endpoints. The model costs $20 per 1M characters on Together Serverless Inference. You choose one of nine preset voices, and a plain-language instruction such as excited and upbeat changes its delivery. First audio arrives in as little as 97ms when you stream the response, over HTTP or WebSocket.
Key details
- Price: $20 per 1M characters on Together Serverless Inference.
- Nine preset voices, and every voice can speak 10 languages: Chinese, English, Japanese, Korean, German, French, Russian, Portuguese, Spanish and Italian.
- First audio in as little as 97ms when streaming.
- Output formats: wav, mp3, raw or mulaw, at a sample rate of 24000 (default) or 8000.
- Word timestamps (alignment=word) are not supported for Qwen3 TTS.
Source
Qwen3 TTS is now available on Together AI. Keep one voice across your product, change its tone with a plain-language instruction and move it between 10 languages, with first audio in as little as 97ms.
Original announcement: Together AI blog
Source: Together AI blog (2026-11-20)