Daily-Omni: Towards Audio-Visual Reasoning with Temporal Alignment across Modalities

Ziwei Zhou¹ Rui Wang¹ Zuxuan Wu^2,*

¹Computation and Artificial Intelligence Innovative Collage, Fudan University

²Institute of Trustworthy Embodied AI, Fudan University

Paper GitHub 🤗 Dataset 🏆 Leaderboard

QA Examples

Examples from our Daily-Omni benchmark, showcasing diverse audio-visual reasoning questions.

Generation Pipeline

The Question-Answer generation pipeline of Daily-Omni.

Annotation Detail showing how we align audio and visual events.

Daily-Omni Agent

Architecture of Daily-Omni Agent.

Model Performance

A case study on the importance of temporal alignment for visual reasoning.

Leaderboard

Methods	AV Event Alignment	Comparative	Context Understanding	Event Sequence	Inference	Reasoning	30s Subset	60s Subset	Avg
Omni-Modal Language Models (With Visual and Audio)
Gemini 2.0 Flash	62.18	73.28	63.73	63.72	76.62	75.43	67.23	68.55	67.84
Daily-Omni (ours)	51.68	68.70	60.10	53.92	78.57	71.43	63.99	59.27	61.82
Gemini 2.0 Flash Lite	55.04	64.89	58.03	54.25	74.03	72.00	62.44	60.00	61.32
Ola (7B)	40.34	61.07	40.41	43.46	63.64	69.71	51.47	49.82	50.71
Qwen2.5-Omni (7B)	44.12	51.15	38.86	40.52	57.79	61.71	46.68	48.36	47.45
Qwen2.5-Omni (3B)	38.66	48.09	33.68	33.99	54.55	44.00	42.35	38.36	40.52
VideoLLaMA2 (7B)	35.71	35.88	35.75	31.70	40.91	34.29	38.02	31.82	35.17
Unified-IO-2 XL (3B)	30.25	30.53	25.39	29.08	33.12	21.71	28.13	28.55	28.32
Unified-IO-2 XXL (8B)	25.63	31.30	26.42	25.82	35.06	29.71	26.74	30.00	28.24
Unified-IO-2 L (1B)	27.31	22.90	26.42	27.78	29.87	29.14	27.67	27.09	27.40

Citation


@misc{zhou2025dailyomni,
      title={Daily-Omni: Towards Audio-Visual Reasoning with Temporal Alignment across Modalities}, 
      author={Ziwei Zhou and Rui Wang and Zuxuan Wu},
      year={2025},
      eprint={2505.17862},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2505.17862}, 
}