Video Question Answering on LVBench Avg
52.5Average AccuracyEFlow
Evaluation results
| Method | Links | |
|---|---|---|
| EFlowFramework Type=Native Multi-turn Tool Invocation Video MLLM, Open-source status=Open-source2026.07 | 52.5 | |
| Qwen3-VL*Framework Type=Single-Turn Video MLLM, Open-source status=Open-source2026.07 | 50.6 | |
| Qwen2.5-VLFramework Type=Single-Turn Video MLLM, Open-source status=Open-source2026.07 | 45.3 | |
| Rewatch-R1Framework Type=Single-Turn Video MLLM, Open-source status=Open-source2026.07 | 43.3 | |
| LLaVA-VideoFramework Type=Single-Turn Video MLLM, Open-source status=Open-source2026.07 | 41.5 | |
| Video-ZoomerFramework Type=Native Multi-turn Tool Invocation Video MLLM, Open-source status=Open-source2026.07 | 41.5 | |
| LongVTFramework Type=Native Multi-turn Tool Invocation Video MLLM, Open-source status=Open-source2026.07 | 41.3 | |
| ConanFramework Type=Native Multi-turn Tool Invocation Video MLLM, Open-source status=Open-source2026.07 | 39.2 | |
| Video-ThinkerFramework Type=Single-Turn Video MLLM, Open-source status=Open-source2026.07 | 37 | |
| VideoAgentFramework Type=Agent-based Framework, Open-source status=Open-source2026.07 | 29.3 | |
| VideoTreeFramework Type=Agent-based Framework, Open-source status=Open-source2026.07 | 28.8 |