List your company

Multi-choice Video Question Answering on VideoMME

75Accuracy (no subs)Gemini-1.5-Pro
40.99249.82158.6567.479Mar 24, 2025Apr 12, 2025May 1, 2025May 20, 2025Jun 8, 2025Jun 27, 2025Jul 17, 2025

Evaluation results

MethodLinks
Gemini-1.5-Prozero-shot=true2025.037581.3—
GPT-4ozero-shot=true2025.0371.977.2—
Gemini-1.5-Flashzero-shot=true2025.0370.375—
InternVL2.5-8B-ITG-32Frames Sampled=322025.0767.369.6—
Qwen2.5-VL-7B2025.0765.171.6—
InternVL2.5-8B2025.0764.266.9—
LLaVAction-7Bzero-shot=true2025.0363.971.4—
LLaVA-Video-7Bzero-shot=true2025.0363.369.7—
LLaVA-Video-7B2025.0763.369.7—
LongVU-7B2025.0760.6——
VILA-40B2025.0760.161.1—
GPT-4Vzero-shot=true2025.0359.963.3—
mPLUG-Owl3zero-shot=true2025.0359.368.1—
LLaVA-OV-7Bzero-shot=true2025.0358.261.5—
LLaVA-OV-7B2025.0758.261.5—
mPLUG-Owl3-8B2025.0753.5——
LongVA-7Bzero-shot=true2025.0352.654.3—
LongVA-7B2025.0752.654.3—
VideoLLaMA2-7Bzero-shot=true2025.0347.950.3—
VideoLLaMA2-7B2025.0745.146.6—
VideoChat2-7Bzero-shot=true2025.0342.354.6—
DYTOLLM Size=7B, Vis Encoder=CLIP-L, Training Regime=Training-Free2026.02——41.2
DYTOLLM Size=34B, Vis Encoder=CLIP-L, Training Regime=Training-Free2026.02——53.4
IG-VLMLLM Size=7B, Vis Encoder=CLIP-L, Training Regime=Training-Free2026.02——39.7
IG-VLMLLM Size=34B, Vis Encoder=CLIP-L, Training Regime=Training-Free2026.02——50.3
KTV-34B-denseLLM Size=34B, Vis Encoder=CLIP-L, Training Regime=Training-Free, Variant=dense2026.02——53.2
KTV-34B-normalLLM Size=34B, Vis Encoder=CLIP-L, Training Regime=Training-Free, Variant=normal2026.02——53
KTV-34B-sparseLLM Size=34B, Vis Encoder=CLIP-L, Training Regime=Training-Free, Variant=sparse2026.02——52.2
KTV-7B-denseLLM Size=7B, Vis Encoder=CLIP-L, Training Regime=Training-Free, Variant=dense2026.02——44
KTV-7B-normalLLM Size=7B, Vis Encoder=CLIP-L, Training Regime=Training-Free, Variant=normal2026.02——43.7
KTV-7B-sparseLLM Size=7B, Vis Encoder=CLIP-L, Training Regime=Training-Free, Variant=sparse2026.02——43.6
SF-LLaVA-7BLLM Size=7B, Vis Encoder=CLIP-L, Training Regime=Training-Free2026.02——39.4
SF-LLaVA-7BLLM Size=34B, Vis Encoder=CLIP-L, Training Regime=Training-Free2026.02——51.9
Video-LLaVALLM Size=7B, Vis Encoder=ViT-L, Training Regime=Training-Based2026.02——39.9