OmniVideo-R1: Reinforcing Audio-visual Reasoning with Query Intention and Modality Attention
Open paper
Benchmark
Model
Rank
Results
audio-visual-question-answering-on-dailyomni
OmniVideo-R1 (Qwen3-Omni-30B-A3B)
–
Accuracy: 82.8
video-question-answering-on-lvbench
OmniVideo-R1 (Qwen3-Omni-30B-A3B)
–
Accuracy: 51.9
Rank counts only results with a code link.