SenseVoiceLarge release updates?

[as4193]

I've been looking at SenseVoiceLarge benchmarks and have been impressed by how its outperforming other models. As such, I'm really excited about its potential applications in the field of ASR. But I noticed that only the Small model is available. Does anyone have any updates on when SensevVoiceLarge is expected to be released ? I'm eager to start exploring testing out and would love to hear any information you might have.

[LauraGPT]

Update as of 2026-07-15: an official SenseVoiceLarge checkpoint has not been publicly released, and there is no published release ETA that we can reliably provide.

I rechecked the current official release surfaces:

• The SenseVoice model zoo links only SenseVoiceSmall on ModelScope and Hugging Face.
• The FunAudioLLM Hugging Face organization currently exposes SenseVoiceSmall and its GGUF conversion, but no Large checkpoint.
• A complete listing of the 1,065 public models under ModelScope's official iic group contains SenseVoiceSmall and SenseVoiceSmall-onnx, but no Large checkpoint.
• The repository releases contain the SenseVoice v1.0.0 release and llama.cpp runtime releases, but no SenseVoiceLarge weights.

The SenseVoiceLarge figures in the paper/benchmark describe the research model used for those experiments; they should not be interpreted as a downloadable checkpoint or a release-date commitment.

For work that can start now:

• Use iic/SenseVoiceSmall or FunAudioLLM/SenseVoiceSmall for the released five-language ASR, language ID, emotion, and audio-event path.
• For CPU/edge deployment, use FunAudioLLM/SenseVoiceSmall-GGUF.
• If the main requirement is higher-capacity ASR rather than SenseVoice's emotion/event outputs, evaluate the separate Fun-ASR-Nano-2512 family. It is not a drop-in SenseVoiceLarge replacement and has a different capability contract.

Please treat the official model zoo and repository releases as the canonical sources. Until a maintainer publishes a checkpoint and model card there, any date would be speculation.