Open-ended Question-Answering on Vad-Reasoning-Plus benchmark leaderboard · SOTA2 Research

Open-ended Question-Answering on Vad-Reasoning-Plus

0.106BLEU-3Qwen3VL-Thinking

Evaluation Results

Method	Links
Qwen3VL-Thinking2026.01		0.106	0.008	0.044
Vad-R1-Plus2026.01		0.03	0.019	0.078
Vad-R12026.01		0.023	0.013	0.058
Gemini2.5-Flash2026.01		0.019	0.01	0.05
LLaVA-OneVision2026.01		0.017	0.009	0.049
Grok4-Fast2026.01		0.015	0.009	0.043
Holmes-VAU2026.01		0.014	0.008	0.045
Kimi-VL-Thinking2026.01		0.013	0.007	0.038
Qwen3VL-Instruct2026.01		0.012	0.006	0.036
Kimi-VL-Instruct2026.01		0.011	0.006	0.035
OpenR1-Video2026.01		0.011	0.006	0.039
VideoChat-R12026.01		0.011	0.006	0.037
MiniCPM-V4.52026.01		0.011	0.005	0.037
InternVL3.52026.01		0.009	0.005	0.031
Ovis2.52026.01		0.009	0.004	0.032
VideoChat-Flash2026.01		0.008	0.004	0.029
GPT5-mini2026.01		0.005	0.002	0.026
Video-XL2026.01		0.004	0.002	0.013
Video-R12026.01		0.004	0.002	0.01
Keye-VL1.52026.01		0.004	0.002	0.022
MiMo-VL2026.01		0.004	0.002	0.019
HAWK2026.01		0.004	0.002	0.019
VideoLLaMA32026.01		0.003	0.002	0.014
InternVideo2.52026.01		0.002	0.001	0.01
LLaVA-NeXT-Video2026.01		0.002	0.001	0.009
Holmes-VAD2026.01		0.002	0.001	0.009
GLM4.1-V2026.01		0.001	0	0.008