Long-Video Question Answering on LVBench

74.2All Score

Deep Video Discovery

Updated 1mo ago

Evaluation Results

Method
Deep Video Discovery 2026.06	74.2	-	-	-	-
GOPAgen 2026.06	73.5	-	-	-	-
MR. Video 2026.06	60.8	-	-	-	-
VideoLucy 2026.06	58.8	-	-	-	-
Qwen3-VL 2026.04	58	-	-	-	-
OpenAI o3 2026.06	57.1	-	-	-	-
AdaReTaKe 2026.06	53.3	-	-	-	-
Qwen3-Omni 2026.04	50.2	-	-	-	-
GPT-4o 2026.06	48.9	-	-	-	-
Gemini-2.0-Flash 2026.06	48.3	-	-	-	-
Omni-o3 2026.04	48.1	-	-	-	-
Qwen2.5-VL-72B 2026.06	47.7	-	-	-	-
Qwen2.5-VL 2026.04	45.3	-	-	-	-
Open-o3-video 2026.04	44.5	-	-	-	-
InternVL2.5-78B 2026.06	43.6	-	-	-	-
mPLUG-Owl3 2026.06	43.5	-	-	-	-
HumanOmniV2 2026.04	43.2	-	-	-	-
Qwen2.5-Omni 2026.04	41.5	-	-	-	-
VCA 2026.06	41.3	-	-	-	-
VideoChat-R1 2026.04	40.8	-	-	-	-
VideoRFT 2026.04	40.7	-	-	-	-
EchoInk-R1 2026.04	38.7	-	-	-	-
AVATAR 2026.04	38.4	-	-	-	-
Gemini1.5Pro 2026.04	33.1	-	-	-	-
Gemini-1.5-Pro 2026.06	33.1	-	-	-	-
Omni-R1 2026.04	32.9	-	-	-	-
GPT-4o 2026.04	30.8	-	-	-	-
Ola 2026.04	29.8	-	-	-	-
AV-Reasoner 2026.04	29.4	-	-	-	-
VideoAgent 2026.06	29.3	-	-	-	-
VideoTree 2026.06	28.8	-	-	-	-
ClipRAG 2025.04	-	33.72	-	-	-
UniversalRAG (Qwen3-VL-2B-Instruct) 2025.04	-	33.72	-	-	-
Oracle 2025.04	-	33.72	-	-	-
Local-Only (LO) 2026.06	-	32.42	0	-	2.72
Cloud-Only (CO) 2026.06	-	39.23	2,549.78	-	4.08
Edge-Cloud Hybrid (EC) 2026.06	-	36.59	637.45	75	4.09
Co-VStream 2026.06	-	38.93	316.42	87.59	2.99