Alibaba's Qwen-Audio-3.0-TTS Plus tops the Speech Arena leaderboard

Alibaba Tongyi Lab released Qwen-Audio-3.0-TTS on July 20, 2026 in Flash and Plus versions for 16 languages and 20 Chinese dialects. Plus placed first on Artificial Analysis Speech Arena at about 1,236 Elo. It generates up to three minutes of audio at up to 48 kHz and supports 86 inline cues including laughter, breathing and sighs. Alibaba Cloud Model Studio hosts it at $27.59 per million characters.

Visit original source