Zen3 ASR

Canonical Zen3 automatic speech recognition model for high-accuracy transcription, voice agents, and real-time streaming workloads. The 1.7B parameter flagship of the Zen3 audio family.

Hosted via Hanzo gateway

Served at api.hanzo.ai as zen3-asr.

Weights

Mirror staging in progress. Use the canonical base via the base_model link in frontmatter until weights land here.

Acknowledgements

Built on Qwen/Qwen3-ASR-1.7B (Apache-2.0). Mirrored here for the Zen3 audio family.

Downloads last month: 25

Safetensors

Model size

2B params

Tensor type

BF16

Model tree for zenlm/zen-3-asr

Base model

Qwen/Qwen3-ASR-1.7B

Finetuned

(61)

this model

Collection including zenlm/zen-3-asr

Zen3 Audio

Collection

Speech recognition + text-to-speech. • 7 items • Updated 2 days ago