Pyyan / Compare / Voxtral Mini TTS 3 vs Qwen-Audio-3.0-TTS-Plus
Voxtral Mini TTS 3 vs Qwen-Audio-3.0-TTS-Plus
2 of 5Voice & Audio · verified 26 Oct 2026
Voxtral Mini TTS 3Mistralcurrent Qwen-Audio-3.0-TTS-PlusAlibabacurrent | Specification | Voxtral Mini TTS 3 | Qwen-Audio-3.0-TTS-Plus |
|---|
| Summary | A frontier lab's first serious text to speech, priced well under the incumbents. | Statistically tied for first. |
|---|
| Arena Elo | Not yet rated | 1230 |
|---|
| Price ($/M chars) | $13 | — |
|---|
| Clone from | ~30 min for Pro Voice | — |
|---|
| Open weights | Not stated | — |
|---|
| Category | Voice & Audio | Voice & Audio |
|---|
| Official | Mistral ↗ | Alibaba ↗ |
|---|
Highlighted rows are where these differ.
Voxtral Mini TTS 3
- 35+ languages and 50+ dialects, in batch and realtime modes
- Pro Voice adapts to a speaker from about thirty minutes of audio
- Ships with a LiveKit voice agent SDK that completes a turn in under a second over SIP or WebRTC
- Mistral has not stated a licence, so whether the weights are open is not yet knowable
Best for multilingual synthesis inside a voice agent.
Full spec sheet →
Qwen-Audio-3.0-TTS-Plus
- Superseded on 23 September by Qwen-Audio-3.1, a five model line including TTS-Next and ASR-Next
- Alibaba cut voice API prices with it: text to speech about 70%, realtime about 85%, speech recognition up to 95%
- No per unit rate published for the new line yet
Best for multilingual.
Full spec sheet →