TencentHy-SimulMT
Hy-SimulMT-30B-A3B
Hy-SimulMT-30B-A3B supports incremental translation between Chinese and English. The October 2026 release processes incoming speech while the speaker continues.
Model design and access
The project supplies SimulInfer serving and SimulTrain training tools. The full demonstration produces speech and can preserve voice characteristics, but speech output and voice cloning require a configured TTS service. A separate Interleave checkpoint is also published.Hy-SimulMT-30B-A3B official model card
Practical use
Practical guidance. Measure translation lag and revisions separately from time to first audio. Use complete meetings with pauses, corrections and interrupted sentences. Connect a reviewed speech service if your application needs spoken output.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | Downloadable |
| License | Apache 2.0Hy-SimulMT-30B-A3B official model card |
| Language directions | Chinese to English and English to ChineseHy-SimulMT-30B-A3B official model card |