# Ming-3.0-audio-flash Benchmark Scores & Performance

> Ming-3.0-audio-flash has inclusionAI Ming-3.0 results results. Voice evidence is display-only and does not produce a text-model score.

Canonical page: https://benchlm.ai/models/ming-3-0-audio-flash

Last updated: October 20, 2026

## Model Details

| Property | Value |
|----------|-------|
| Creator | InclusionAI |
| Source Type | Open Weight |
| Reasoning Type | Non-Reasoning |
| Context Window | N/A |
| Official model card | [InclusionAI model documentation](https://huggingface.co/inclusionAI/Ming-3.0-audio-flash) |
| Overall Score | Not scored (voice evidence only) |
| Overall Rank | Unranked |

## Family & Coverage

- Family: Ming-3.0-audio-flash
- Variant: base
- Benchmarks covered: 0 of 670
- Coverage note: Speech and voice results appear below; no weighted text-model benchmark rows are stored.

## Speech and voice evidence

[Ming-3.0 speech evaluation](/voice-benchmarks/ming-3-0-speech-evaluation)

All results are inclusionAI-run or developer-reported by inclusionAI. Text-to-speech figures use a unified ASR-based WER or CER protocol over each open model at its recommended settings; speaker similarity (SIM) trails content accuracy, and Korean CER for Ming-3.0-tts-tiny (6.000%) is the clearest multilingual gap. The audio and ASR figures are read from images on the model cards; results marked with an asterisk there are self-evaluated. Competitor rows are not carried here. These results do not enter text-model or voice-agent rankings.

| Metric | Result |
| --- | --- |
| Chinese ASR average CER | 3.45 |
| English ASR average WER | 3.69 |
| Dialect ASR average CER | 9.40 |
| BigBenchAudio score | 99.40 |
| VoiceBench overall | 86.74 |

inclusionAI model cards retrieved October 21, 2026. Provider-reported; error rates are better lower. Voice results are display-only.

- [Evaluation source](https://huggingface.co/inclusionAI/Ming-3.0-tts-tiny)

MIT-licensed weights on Hugging Face and ModelScope, first uploaded September 3, 2026 and refreshed October 21, 2026. The card gives 126B total and 6.5B activated parameters; the checkpoint stores 128B. It handles Chinese and English with accents and dialects, takes audio in and returns text or text plus synthesized speech, and documents a 128K reference serving profile on eight 80 GB H800 GPUs.

## Other InclusionAI Models

- [Ling 3.1 Flash](/models/ling-3-1-flash) - Score: 68.05
- [Ling 3.0 Flash VL](/models/ling-3-0-flash-vl) - Score: 46.79
- [Ling 3.0 Flash](/models/ling-3-0-flash) - Score: 45.06
- [Ling 2.6 Flash](/models/ling-2-6-flash) - Score: 32.43
- [Ling 3.0 Flash Sante](/models/ling-3-0-flash-sante) - Score: not computed
- [Ling 3.0 Flash FP8](/models/ling-3-0-flash-fp8) - Score: not computed
- [Ling 3.0 Tiny](/models/ling-3-0-tiny) - Score: not computed
- [LLaDA2.2-flash](/models/llada2-2-flash) - Score: not computed
- [LLaDA2.2-mini](/models/llada2-2-mini) - Score: not computed
- [Ling 3.0 Flash Fin](/models/ling-3-0-flash-fin) - Score: not computed
- [Ming-3.0-asr-tiny](/models/ming-3-0-asr-tiny) - Score: not computed
- [Ming-3.0-audio-tiny](/models/ming-3-0-audio-tiny) - Score: not computed
- [Ming-3.0-tts-0.5B](/models/ming-3-0-tts-0-5b) - Score: not computed
- [Ming-3.0-tts-tiny](/models/ming-3-0-tts-tiny) - Score: not computed