Visual Question Answering on CameraBench benchmark leaderboard · SOTA2 Research

Visual Question Answering on CameraBench

57.6Motion Steadiness AccuracyOpen-o3-Video-7B

Evaluation Results

Method	Links
Open-o3-Video-7B2025.10		57.6	—	69.3	—	51.3	—	—	—	—	58.8
Qwen2.5-VL-7B2025.10		56.7	—	69	—	49.3	—	—	—	—	57.5
Open-o3-Video-7B2025.10		56.7	—	68.7	—	50	—	—	—	—	58.5
Open-o3-Video-7B2025.10		55.9	—	67	—	50.3	—	—	—	—	57.8
CamReasoner-7B2026.01		0.78	0.77	0.741	0.674	0.607	0.718	0.713	0.606	0.826	0.745
Gemini-2.5-Pro2026.01		0.596	0.525	0.615	0.501	0.468	0.536	0.509	0.439	0.661	0.561
Qwen2.5-VL-3B2026.01		0.593	0.583	0.606	0.54	0.517	0.557	0.576	0.5	0.667	0.592
Qwen2.5-VL-72B2026.01		0.577	0.621	0.71	0.578	0.532	—	0.592	0.504	0.686	0.603
InternVL2.5-26B2026.01		0.576	0.65	0.623	0.551	0.524	0.618	0.598	0.538	0.549	0.586
InternVL3-78B2026.01		0.576	0.65	0.623	0.551	0.524	0.618	0.598	0.538	0.549	0.586
GPT-4o2026.01		0.572	0.539	0.627	0.595	0.546	0.656	0.53	0.431	0.634	0.604
Qwen2.5-VL-7B2026.01		0.568	0.606	0.629	0.569	0.54	0.584	0.642	0.536	0.644	0.603
InternVL2.5-8B2026.01		0.558	0.612	0.589	0.526	0.509	0.595	0.565	0.512	0.512	0.558
InternVL3-8B2026.01		0.558	0.612	0.589	0.526	0.509	0.595	0.565	0.512	0.512	0.558
LLaVA-OneVision-7B2026.01		0.557	0.653	0.706	0.544	0.568	0.624	0.622	0.444	0.536	0.585
Gemini-2-Flash2026.01		0.549	0.479	0.58	0.456	0.421	0.476	0.476	0.402	0.653	0.531
LLaVA-Video-7B2026.01		0.548	0.677	0.586	0.534	0.511	0.563	0.613	0.526	0.696	0.602
InternVideo2-Chat-8B2026.01		0.537	0.659	0.53	0.515	0.51	0.535	0.497	0.522	0.518	0.526
mPLUG-Owl3-7B2026.01		0.531	0.664	0.629	0.498	0.504	0.554	0.545	0.471	0.649	0.572
Tarsier-Recap-7B2026.01		0.531	0.643	0.517	0.51	0.502	0.528	0.49	0.514	0.51	0.518
Random Chance2026.01		0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5	0.5