ai-sage/GigaChat3.1-Audio-10B-A1.8B · Hugging Face

GigaChat Audio 10B is an audio-native LLM built on top of the GigaChat 3.1 Lightning text model. A Conformer speech encoder and a modality adapter feed audio embeddings directly into a Mixture-of-Experts decoder, so the model keeps the text quality of its base while adding speech understanding.

Capabilities: audio question answering and classification, temporal grounding (localization in long audio, timestamped event descriptions, audio summarization with timestamps), tool-use, and text-only tasks.

The temporal grounding skills are trained on TimeGround-1M — a purpose-built dataset of long-form audio paired with time-aligned annotations.

• arXiv : https://arxiv.org/abs/2607.10387
• Full Paper : https://arxiv.org/pdf/2607.10387.pdf
• HF Dataset : https://huggingface.co/datasets/ai-sage/TimeGround-1M
• HF Space : https://huggingface.co/spaces/hugging-apps/gigachat-audio-10b-a1-8b-demo