Long Video Question Answering on Video-MME w/o subtitles
0.818AccuracyGPT-5
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5LLM=N/A, # Frames=N/A2025.12 | 0.818 | |
| Gemini-1.5-ProLLM=N/A, # Frames=N/A2025.12 | 0.75 | |
| GPT-4oLLM=N/A, # Frames=N/A2025.12 | 0.719 | |
| InternVL3.5LLM=8B, # Frames=642025.12 | 0.66 | |
| AKSLLM=7B, # Frames=642025.12 | 0.653 | |
| Qwen2.5-VLLLM=7B, # Frames=322025.12 | 0.651 | |
| LLaVA-Video + OneClip-RAGLLM=7B, # Frames=642025.12 | 0.649 | |
| ByteVideoLLMLLM=14B, # Frames=2562025.12 | 0.646 | |
| NVILALLM=8B, # Frames=2562025.12 | 0.642 | |
| LLaVA-VideoLLM=7B, # Frames=642025.12 | 0.633 | |
| LongVULLM=7B, # Frames=1fps2025.12 | 0.606 | |
| mPLUG-Owl3LLM=7B, # Frames=162025.12 | 0.593 | |
| Video-XLLLM=7B, # Frames=1282025.12 | 0.555 | |
| LongVALLM=7B, # Frames=2562025.12 | 0.526 |