List your company

Audio-Visual Question Answering on Video Holmes 32 frames

64.4SRStreamOV
33.40841.45449.557.546May 25, 2026

Evaluation results

MethodLinks
StreamOVAudio=true2026.0564.456.252.444.548.553.650.453.1
Qwen3-Omni-30BAudio=true2026.0560.662.348.435.550.95155.252.8
Gemini 2.0 FlashAudio=true2026.0556.554.243.444.543.955.150.149.5
GPT-4oAudio=false2026.055049.638.8304439.23742
Qwen3-VL-30BAudio=false2026.0546.638.435.53939.835.135.638.8
Gemini 2.5 FlashAudio=true2026.0543.446.943.15137.943.639.343.1
Qwen2.5-Omni-7BAudio=true2026.0541.835.123.42528.917.525.929
Qwen2.5-VL-7BAudio=false2026.0534.621.415.412.517.212.415.219