General Video Understanding on LVBench
53.8AccuracyGLM-4.5V
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GLM-4.5V2026.04 | 53.8 | — | |
| QCALLM Size=30B, Frames=64, Backbone=Qwen3-VL-30B-A3B2026.07 | 52.6 | — | |
| QCALLM Size=8B, Frames=64, Backbone=Qwen3-VL-8B2026.07 | 51.8 | — | |
| GPT-4o2026.04 | 48.9 | — | |
| GPT-4oEvaluation Protocol=vanilla offline2026.05 | 48.9 | — | |
| Qwen2.5-VL-72BLLM Size=72B, Frames=1fps2026.07 | 47.3 | — | |
| LLaVA-Video-72BLLM Size=72B, Frames=642026.07 | 45.5 | — | |
| Video-LLaMA3-7BEvaluation Protocol=vanilla offline2026.05 | 45.3 | — | |
| Qwen3-VL-30B-A3BLLM Size=30B, Frames=642026.07 | 44 | — | |
| Qwen3-VL-8BLLM Size=8B, Frames=642026.07 | 43.8 | — | |
| mPLUG-Owl3LLM Size=8B, Frames=1282026.07 | 43.5 | — | |
| InternVL-V2.5-8BEvaluation Protocol=vanilla offline2026.05 | 43.2 | — | |
| InternVL-V3.5-8BEvaluation Protocol=vanilla offline2026.05 | 42.5 | — | |
| EvoStreaming-8B-InternVL-V3.5Evaluation Protocol=vanilla offline2026.05 | 42.2 | — | |
| Flash-VStream-7BEvaluation Protocol=vanilla offline2026.05 | 42 | — | |
| LLaVA-Video-7BEvaluation Protocol=vanilla offline2026.05 | 41.5 | — | |
| Qwen2.5-VL-7BEvaluation Protocol=vanilla offline2026.05 | 40.2 | — | |
| SFTDistillation Method=SFT, Teacher Model=Qwen3-235B2026.04 | 39.9 | — | |
| EvoStreaming-8B-Qwen2.5Evaluation Protocol=vanilla offline2026.05 | 39.8 | — | |
| Qwen2-VL-7BEvaluation Protocol=vanilla offline2026.05 | 39.1 | — | |
| EvoStreaming-8B-Qwen2Evaluation Protocol=vanilla offline2026.05 | 38.8 | — | |
| AVATAR2026.04 | 38.4 | — | |
| Kangaroo-7BEvaluation Protocol=vanilla offline2026.05 | 38.3 | — | |
| FFRBase Model=Video-R1-SFT, Training=RFT + FFR2026.04 | 38.1 | 24.1 | |
| FFR (GLM-4.5V)Distillation Method=FFR, Teacher Model=GLM-4.5V2026.04 | 38.1 | — | |
| FFRBase Model=VideoRFT-SFT, Training=RFT + FFR2026.04 | 37.8 | 40.52 | |
| Video-Thinker2026.04 | 37 | — | |
| SFTDistillation Method=SFT, Teacher Model=Qwen3-32B2026.04 | 36.7 | — | |
| Video-R12026.04 | 35.3 | — | |
| Video-ChatR12026.04 | 34.3 | — | |
| Pixel-Reasoner2026.04 | 34.3 | — | |
| FFRDistillation Method=FFR, Teacher Model=Qwen3-32B2026.04 | 34.2 | — | |
| VideoRFT2026.04 | 33.9 | — | |
| FFRDistillation Method=FFR, Teacher Model=Qwen3-235B2026.04 | 33.9 | — | |
| Gemini-1.5-proEvaluation Protocol=vanilla offline2026.05 | 33.1 | — | |
| Gemini-1.5-ProLLM Size=-, Frames=-2026.07 | 33.1 | — | |
| Qwen2.5-VL2026.04 | 31.6 | — | |
| Video-R1-SFT2026.04 | 30.7 | — | |
| Video-R1-SFTDistillation Method=SFT, Teacher Model=Baseline2026.04 | 30.7 | — | |
| VideoRFT-SFT2026.04 | 26.9 | — | |
| VideoLLM-online-8BEvaluation Protocol=vanilla offline2026.05 | 24 | — | |
| TimeChatEvaluation Protocol=vanilla offline2026.05 | 22.3 | — |