List your company

Video Understanding on MVBench zero-shot

69.7AccuracyVideoLLaMA3
31.2241.2151.261.19Jul 29, 2025Jul 30, 2025Aug 1, 2025Aug 3, 2025Aug 4, 2025Aug 6, 2025Aug 8, 2025

Evaluation results

MethodLinks
VideoLLaMA3LLM=Qwen2.5-7B, Frames=1fps2025.0769.7—————————
Qwen2.5-VLLLM=Qwen2.5-7B2025.0769.6—————————
InternVL3.5-REGATELLM=Qwen3-14B, Frames=16, Tokens=2.32B (↓ 41.41%)2025.0769.6—————————
InternVL3.5LLM=Qwen3-14B, Frames=16, Tokens=3.96B2025.0768.3—————————
Qwen-2.5-VLSize=3B, #V=31932025.0865.268.367.748.891637551.575.556
GPT-4o2025.0764.6—————————
Fourier-Qwen-2.5Size=3B, #V=13282025.0862.967.265.748.590.554.870.547.56657.5
Qwen-2-VLSize=2B, #V=31932025.0861.468.565.744.890.560.865.353.55947.8
Gemini-1.5-Pro2025.0760.5—————————
Fourier-Qwen-2Size=2B, #V=13282025.0859.867.763.34088.555.564.8515750.2
LLaVA-OneVisionLLM=Qwen2-7B, Frames=322025.0756.7—————————
VideoChat2-REGATELLM=Mistral-7B, Frames=16, Tokens=2.22B (↓ 43.51%)2025.0756.6—————————
VideoChat2LLM=Mistral-7B, Frames=16, Tokens=3.93B2025.0755.7—————————
VideoLLaMA2-REGATELLM=Qwen2-7B, Frames=16, Tokens=49.27M (↓ 41.22%)2025.0753.6—————————
VideoLLaMA2LLM=Qwen2-7B, Frames=16, Tokens=83.82M2025.0752—————————
VideoChat2Size=7B, #V=40962025.0851.161.357.32388.540.551.34936.547
LLaVA-NeXT-VideoLLM=Vicuna-7B, Frames=322025.0746.5—————————
LLaVA-v1.5Size=7B, #V=23042025.0845.652.843.731.583.53745.344.55037.3
Fourier-LLaVASize=7B, #V=2882025.084448.146.3318140.541.3364936.3
Video-LLaVASize=7B, #V=40962025.0843.14846.527.884.535.545.83442.534.2
LLaMA-VIDLLM=Llama-2-7B, Frames=1fps2025.0741.9—————————
Video-LLaVALLM=Vicuna-7B, Frames=82025.0741—————————
VideoChatSize=7B, #V=40962025.0835.53841.226.348.527.844.326.54127.7
Video-LLaMASize=7B, #V=402025.0834.134.442.222.54328.33932.54029.3
VideoChatGPTSize=7B, #V=3562025.0832.732.140.721.5312844293330.3