Is Ming-3.0 state of the art?Pending benchmarks.Announced Oct 19, 2026; no independent results yet.

The details so far

Lab

Ant Ling (InclusionAI, Ant Group)

Announced

Oct 19, 2026

API

Not yet

Open weights

Published

Parameters

0.5B dense TTS; TTS-tiny 7.9B/1.3B MoE; audio-tiny 9.3B/2.7B MoE; audio-flash 126B/6.5B MoE

Models

Ming-3.0-tts-0.5B (dense), Ming-3.0-tts-tiny (7.9B/1.3B MoE), Ming-3.0-asr-tiny, Ming-3.0-audio-tiny (9.3B/2.7B MoE), Ming-3.0-audio-flash (126B/6.5B MoE)

TTS

Voice cloning and voice design; Chinese, English, German, Spanish, French, Korean, Japanese; Sichuanese and Cantonese dialects on the TTS cards

ASR

Ming-3.0-asr-tiny: Chinese and English transcription with regional dialects and optional hotword/context prompts

Audio-LM

audio-flash and audio-tiny accept audio plus text; ASR, spoken QA, audio understanding/reasoning, optional speech responses

License

MIT on the Hugging Face cards

Weights

Hugging Face under inclusionAI; also listed on ModelScope

Tags marked unsure are not stated on the lab's own pages.

What would let us call it

Verdicts use only independent boards we already track. Two consecutive snapshots are required before any crown.

• Publicly announced with a source
• Usable by the public (API or weights)
• Listed on an independent board we track
• Enough results for a rank in a category
• Holds across two snapshots (crowns only)

What the lab claims

Self-reported. Shown for context; not used in any verdict.

ModelScope post claim: Ming-3.0-tts-0.5B is open-weight SOTA on Artificial Analysis Controlled Voice at 1069 Elo (self-reported in that post).

Sources