Ant's Ming-3.0-tts hits top-tier TTS quality with just 0.5B params, MIT licensed

ArtificialAnlys · x · 2026-10-23

Per Artificial Analysis, Ant Group's Ming-3.0-tts delivers top-tier text-to-speech quality with only 0.5B parameters under an MIT license. It is the only sub-1B model in the top 3 of any of their TTS leaderboards — every other open-weights TTS model in the Controlled Voice Arena top 10 is 3.5B or larger, meaning it matches that tier with 7x to 70x fewer parameters.

Related event: Ant Group's 0.5B Ming-3.0-TTS tops TTS rankings(2 posts)→

Original post →

More from Multimodal

Multimodal channel →