Temporal Video Understanding on TempCompass
83.7AccuracySeed1.5-VL
Evaluation results
| Method | Links | |||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Seed1.5-VLThinking mode=true, Evaluation frame rate=1 FPS2025.05 | 83.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Seed1.5-VLThinking mode=false, Evaluation frame rate=1 FPS2025.05 | 83.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gemini 1.5 ProSource=[59]2026.06 | 82.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gemini 1.5 ProSource=[60]2026.06 | 81.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GPT-4o2026.06 | 80.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gemini 1.5 Flash2026.06 | 80.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-8B + CRPOBackbone=Qwen3-VL-8B, Post-training Method=CRPO2026.05 | 77.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Structured Reward Reinforcement LearningBackbone Model=Qwen3-VL-32B, Method/Configuration=+Ours, Model Access Level=Open-source2026.04 | 77.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-32BBackbone Model=Qwen3-VL-32B, Method/Configuration=Base, Model Access Level=Open-source2026.04 | 77.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GPT-5.1Model Type=Proprietary2026.05 | 76.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RLER2026.04 | 76.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AdaCodecToken budget=comparable, Backbone=Qwen2-VL-7B2026.06 | 75.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gemini 2.5 ProEvaluation frame rate=1 FPS, self-collected via API=true2025.05 | 75.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AdaCodecToken budget=1/7, Backbone=Qwen2-VL-7B2026.06 | 75.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Molmo2-8B + MotionAtlas-DataModel Class=Open-source Models, Training Data=MotionAtlas-Data2026.06 | 75.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FFRBase Model=Video-R1-SFT, Training=RFT + FFR2026.04 | 75.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 9.25 | |
| FFR (GLM-4.5V)Distillation Method=FFR, Teacher Model=GLM-4.5V2026.04 | 75.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-8B + GRPOBackbone=Qwen3-VL-8B, Post-training Method=GRPO2026.05 | 75.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Key-VL-1.5-8B2026.06 | 75.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Pixel-Reasoner2026.04 | 75.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FFRBase Model=VideoRFT-SFT, Training=RFT + FFR2026.04 | 75.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 10.23 | |
| Qwen3-VL-8B + T-GRPOBackbone=Qwen3-VL-8B, Post-training Method=T-GRPO2026.05 | 75.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FFRDistillation Method=FFR, Teacher Model=Qwen3-235B2026.04 | 75.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Structured Reward Reinforcement LearningBackbone Model=Qwen3-VL-8B, Method/Configuration=+Ours, Model Access Level=Open-source2026.04 | 75.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SFTDistillation Method=SFT, Teacher Model=Qwen3-235B2026.04 | 75.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-8BBackbone=Qwen3-VL-8B, Post-training Method=base2026.05 | 75.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Molmo2-4B + MotionAtlas-DataModel Class=Open-source Models, Training Data=MotionAtlas-Data2026.06 | 75.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-8B + MotionAtlas-DataModel Class=Open-source Models, Training Data=MotionAtlas-Data2026.06 | 75.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-8B + ArrowRLBackbone=Qwen3-VL-8B, Post-training Method=ArrowRL2026.05 | 74.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GPT-4o mini2026.06 | 74.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VL-72BFrames=7682026.01 | 74.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-8B-Thinking + MVPFrames=642026.01 | 74.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gemini-3.1-ProModel Type=Proprietary2026.05 | 74.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Eagle-X5-7B2026.06 | 74.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EvoVidModel=Qwen3-VL-8B-Instruct, Training Iteration=Iter 12026.05 | 74.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EvoVidModel=Qwen3-VL-8B-Instruct, Training Iteration=Iter 32026.05 | 74.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2-VL-7B2026.06 | 74.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-8B-ThinkingFrames=642026.01 | 74.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-4B + CRPOBackbone=Qwen3-VL-4B, Post-training Method=CRPO2026.05 | 74.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-4B + MotionAtlas-DataModel Class=Open-source Models, Training Data=MotionAtlas-Data2026.06 | 74.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-8BBackbone Model=Qwen3-VL-8B, Method/Configuration=Base, Model Access Level=Open-source2026.04 | 74.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SeasonBackbone Model=Qwen3-VL-32B, Method/Configuration=+Season, Model Access Level=Open-source2026.04 | 74.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gemini 3 ProModel Class=Closed-source Models2026.06 | 74.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VideoChat-R12026.04 | 73.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GPT-4o2025.12 | 73.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GPT-4oBackbone Model=GPT-4o, Method/Configuration=Base, Model Access Level=Closed-source2026.04 | 73.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Structured Reward Reinforcement LearningBackbone Model=Qwen2.5-VL-7B, Method/Configuration=+Ours, Model Access Level=Open-source2026.04 | 73.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GPT-4o2026.04 | 73.75 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SeasonBackbone Model=Qwen2.5-VL-7B, Method/Configuration=+Season, Model Access Level=Open-source2026.04 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VideoRFT2026.04 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VideoRFT2026.04 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EvoVidModel=Qwen3-VL-8B-Instruct, Training Iteration=Iter 22026.05 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gemini 2.5 ProModel Class=Closed-source Models2026.06 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VideoRFTEvaluation Setting=Direct Answer2026.07 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-8B-InstructEvaluation Setting=Direct Answer2026.07 | 73.62 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Base ModelModel=Qwen3-VL-8B-Instruct, Training Iteration=w/o training2026.05 | 73.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DNA-Train-7B2025.12 | 73.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-4B + T-GRPOBackbone=Qwen3-VL-4B, Post-training Method=T-GRPO2026.05 | 73.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TCDBackbone Model=Qwen2.5-VL-7B, Method/Configuration=+TCD, Model Access Level=Open-source2026.04 | 73.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EvoVidModel=Qwen3-VL-8B-Instruct, Training Iteration=Frozen Questioner2026.05 | 73.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Molmo2-8B2026.06 | 73.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DINO-HEALBackbone Model=Qwen2.5-VL-7B, Method/Configuration=+DINO-HEAL, Model Access Level=Open-source2026.04 | 73.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Video-R12026.04 | 73.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Video-R12026.04 | 73.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EvoVidModel=Qwen2.5-VL-7B-Instruct, Training Iteration=Iter 32026.05 | 73.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Video-R1Evaluation Setting=Direct Answer2026.07 | 73.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VideoSearcher-8BEvaluation Setting=Agentic Model2026.07 | 73.14 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EvoVidModel=Qwen2.5-VL-7B-Instruct, Training Iteration=Iter 22026.05 | 73.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Structured Reward Reinforcement LearningBackbone Model=Qwen3-VL-4B, Method/Configuration=+Ours, Model Access Level=Open-source2026.04 | 73 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SFTDistillation Method=SFT, Teacher Model=Qwen3-32B2026.04 | 73 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Molmo2-O-7B2026.06 | 73 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GPT-5.2Model Class=Closed-source Models2026.06 | 73 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VL-7BBackbone Model=Qwen2.5-VL-7B, Method/Configuration=Base, Model Access Level=Open-source2026.04 | 72.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MOSS-ChatV2026.04 | 72.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Video-ChatR12026.04 | 72.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Claude 3.5 Sonnet2026.06 | 72.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MiniCPM-V-2.6-8B2026.06 | 72.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PLM-7B2026.06 | 72.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ArrowRLBackbone Model=Qwen2.5-VL-7B, Method/Configuration=+ArrowRL, Model Access Level=Open-source2026.04 | 72.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EvoVidModel=Qwen2.5-VL-7B-Instruct, Training Iteration=Frozen Questioner2026.05 | 72.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| STAR-R12026.04 | 72.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| InternVL3.5-8B + MVPFrames=642026.01 | 72.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GLM-4V-9B2026.06 | 72.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VideoChat-R1.5Evaluation Setting=Direct Answer2026.07 | 72.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VL2026.04 | 72.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FFRDistillation Method=FFR, Teacher Model=Qwen3-32B2026.04 | 72.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Base ModelModel=Qwen2.5-VL-7B-Instruct, Training Iteration=w/o training2026.05 | 72.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-4B + GRPOBackbone=Qwen3-VL-4B, Post-training Method=GRPO2026.05 | 72.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EvoVidModel=Qwen2.5-VL-7B-Instruct, Training Iteration=Iter 12026.05 | 72.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Molmo2-8BModel Class=Open-source Models2026.06 | 72.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Streamo-7BModel Category=Streamo Framework, Parameters=7B2025.12 | 71.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-8BModel Class=Open-source Models2026.06 | 71.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VL-7BModel Category=Streamo Framework, Parameters=7B2025.12 | 71.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EvoVidModel=Qwen3-VL-4B-Instruct, Training Iteration=Iter 32026.05 | 71.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ArrowRL*Model Type=Open-Source2026.05 | 71.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-4BBackbone=Qwen3-VL-4B, Post-training Method=base2026.05 | 71.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-VL-4B + ArrowRLBackbone=Qwen3-VL-4B, Post-training Method=ArrowRL2026.05 | 71.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VL-7B2025.12 | 71.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VideoCoMEvaluation Setting=Direct Answer2026.07 | 71.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SeasonBackbone Model=Qwen3-VL-8B, Method/Configuration=+Season, Model Access Level=Open-source2026.04 | 71.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — |