Video Question Answering on Video-MME (Short, Medium, Long)
81.8VQA Accuracy (Overall)GPT-5
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| GPT-5LLM=-2026.03 | 81.8 | — | — | — | |
| Gemini-1.5-ProLLM=-2026.03 | 75 | 81.7 | 74.3 | 67.4 | |
| GPT-4oLLM=-2026.03 | 71.9 | 80 | 70.3 | 65.3 | |
| TSPOLLM=7B2026.03 | 65.5 | — | — | 54.7 | |
| Qwen2.5-VLLLM=7B2026.03 | 65.1 | — | — | — | |
| LLaVA-Video + FlexMemLLM=7B2026.03 | 64.7 | 76.3 | 63.3 | 54.4 | |
| ByteVideoLLMLLM=14B2026.03 | 64.6 | 74.4 | 62.9 | 56.4 | |
| LLaVA-VideoLLM=7B2026.03 | 63.2 | 76.1 | 61 | 52.4 | |
| LongVULLM=7B2026.03 | 60.6 | — | — | 59.5 | |
| mPLUG-Owl3LLM=7B2026.03 | 59.3 | 70 | 57.7 | 50.1 | |
| TimeMarkerLLM=8B2026.03 | 57.3 | 71 | 54.4 | 46.4 | |
| Video-XLLLM=7B2026.03 | 55.5 | 62 | 53.2 | 49.2 | |
| LongVALLM=7B2026.03 | 52.6 | 61.1 | 50.4 | 46.2 |