Alibaba releases Qwen-Audio-3.0-TTS with 16 languages and lower pricing
Alibaba Tongyi Lab released Flash and Plus versions of Qwen-Audio-3.0-TTS on July 20, 2026. The 12.5 Hz speech tokenizer supports 16 languages and 20 Chinese dialects. Flash targets real-time synthesis at roughly 300 ms, while Plus prioritizes quality. Both offer 86 inline nonverbal tags and natural-language style controls. The model topped the Artificial Analysis speech leaderboard, and pricing of $27.59 per million characters was reported at about one-third of ElevenLabs and MiniMax pricing.
Visit original source