3D Question Answering on ScanQA
30.3EM (Exact Match)3DRS
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| 3DRSModel category=Generalists, 3D geometric inputs=Yes2026.03 | 30.3 | 104.8 | — | — | |
| Video-3D LLM (Uniform)3D Generalist=true, sampling strategy=uniform sampling2024.11 | 30.1 | 102.1 | — | — | |
| Video-3D LLMModel category=Generalists, 3D geometric inputs=Yes2026.03 | 30.1 | 102.1 | — | — | |
| Video-3D LLMVideo sequence length=322026.04 | 30.1 | — | 102.1 | — | |
| 3D-IDEModel category=Generalists, 3D geometric inputs=No2026.03 | 29.8 | 102.1 | — | — | |
| Video-3D LLM†Video sequence length=32, 3D Encoder=Additional 3D encoder2026.04 | 29.8 | — | 104.3 | — | |
| Video-3D LLM (MC)3D Generalist=true, sampling strategy=maximum coverage2024.11 | 29.5 | 100.5 | — | — | |
| Video-3D LLM*Model category=Generalists, 3D geometric inputs=No2026.03 | 29.5 | 99.7 | — | — | |
| Geo3DPrunerVideo sequence length=32, 3D Encoder=Additional 3D encoder, Reduction ratio=↓60%2026.04 | 29.2 | — | 101.3 | — | |
| Geo3DPrunerVideo sequence length=32, 3D Encoder=Additional 3D encoder, Reduction ratio=↓80%2026.04 | 28.3 | — | 97.9 | — | |
| LLaVA-OV-7BModel Category=Our Base Model2026.03 | 27.6 | — | — | — | |
| VID-LLMModel category=Generalists, 3D geometric inputs=No2026.03 | 27.6 | 101.9 | — | — | |
| SceneLLM3D Generalist=true2024.11 | 27.2 | 80 | — | — | |
| Scene-LLMModel Category=Task-specialized 3D-LMMs2026.03 | 27.2 | — | — | — | |
| SceneLLMModel category=Generalists, 3D geometric inputs=Yes2026.03 | 27.2 | 80 | — | — | |
| LLaVA-3D3D Generalist=true2024.11 | 27 | 91.7 | — | — | |
| BridgeQAModel Category=Specialized Models2026.03 | 27 | — | — | — | |
| LLaVA-3DModel category=Generalists, 3D geometric inputs=Yes2026.03 | 27 | 91.7 | — | — | |
| LLaVA-3D2026.04 | 27 | — | 91.7 | — | |
| Geo3DPrunerVideo sequence length=32, 3D Encoder=Additional 3D encoder, Reduction ratio=↓90%2026.04 | 26.8 | — | 92.1 | — | |
| MovieChatModel Category=Video-LMMs, Backbone=LLaVA-OV-7B2026.03 | 26 | — | — | — | |
| UniVLGModel Category=Task-specialized 3D-LMMs2026.03 | 25.7 | — | — | — | |
| Inst3D-LLMModel category=Generalists, 3D geometric inputs=Yes2026.03 | 24.6 | 88.6 | — | — | |
| LEOModel Category=Task-specialized 3D-LMMs2026.03 | 24.5 | — | — | — | |
| DSPNetModel Category=Specialized Models2026.03 | 23.5 | — | — | — | |
| 3D-VisTA3D Generalist=false2024.11 | 22.4 | 69.6 | — | — | |
| 3D-VisTAModel Category=Specialized Models2026.03 | 22.4 | — | — | — | |
| 3D-VisTAModel category=Specialists2026.03 | 22.4 | 69.6 | — | — | |
| FE-3DGQAModel Category=Task-specialized 3D-LMMs2026.03 | 22.3 | — | — | — | |
| ChatScene3D Generalist=true2024.11 | 21.6 | 87.7 | — | — | |
| ChatSceneModel category=Generalists, 3D geometric inputs=Yes2026.03 | 21.6 | 87.7 | — | — | |
| LEO3D Generalist=true2024.11 | 21.5 | 101.4 | — | — | |
| LEOModel category=Generalists, 3D geometric inputs=Yes2026.03 | 21.5 | 101.4 | — | — | |
| LEO2026.04 | 21.5 | — | 101.4 | — | |
| ScanQA3D Generalist=false2024.11 | 21.1 | 64.9 | — | — | |
| ScanQAModel Category=Specialized Models2026.03 | 21.1 | — | — | — | |
| ScanQAModel category=Specialists2026.03 | 21.1 | 64.9 | — | — | |
| ScanQA2026.04 | 21.1 | — | 64.9 | — | |
| 3D-LLM(BLIP2-flant5)3D Generalist=false2024.11 | 20.5 | 69.4 | — | — | |
| 3D-LLMModel Category=Task-specialized 3D-LMMs2026.03 | 20.5 | — | — | — | |
| 3D-LLM(BLIP2-flant5)Model category=Generalists, 3D geometric inputs=Yes2026.03 | 20.5 | 69.4 | — | — | |
| 3D-LLM (BLIP2-FlanT5)2026.04 | 20.5 | — | 69.4 | — | |
| 3D-LLM(Flamingo)3D Generalist=false2024.11 | 20.4 | 59.2 | — | — | |
| 3D-LLM(Flamingo)Model category=Generalists, 3D geometric inputs=Yes2026.03 | 20.4 | 59.2 | — | — | |
| 3D-LLM (Flamingo)2026.04 | 20.4 | — | 59.2 | — | |
| VideoChat2Model Category=Video-LMMs2026.03 | 19.2 | — | — | — | |
| LLaVA-NeXT-VideoModel Category=Video-LMMs2026.03 | 18.7 | — | — | — | |
| VoteNet+MCANModel Category=Specialized Models2026.03 | 17.3 | — | — | — | |
| 3D-LLaVAModel category=Generalists, 3D geometric inputs=Yes2026.03 | — | 92.6 | — | — | |
| ACE-Brain 0-8B2026.07 | — | — | — | 97.3 | |
| ACE-Brain 0.5-8B2026.07 | — | — | — | 99.2 | |
| Chat-3D3D Generalist=false2024.11 | — | 53.2 | — | — | |
| Chat-3DModel category=Generalists, 3D geometric inputs=Yes2026.03 | — | 53.2 | — | — | |
| Chat-3D v23D Generalist=true2024.11 | — | 87.6 | — | — | |
| Chat-3D v2Model category=Generalists, 3D geometric inputs=Yes2026.03 | — | 87.6 | — | — | |
| Chat-Scene++Input Modality=2D-only2026.03 | — | — | 85.6 | — | |
| Claude Sonnet 4.62026.07 | — | — | — | 31.5 | |
| Cosmos3 Nano2026.07 | — | — | — | 60.9 | |
| Embodied R1.52026.07 | — | — | — | 67.1 | |
| Gemini 2.5 Pro2026.07 | — | — | — | 67 | |
| Gemini-1.5-FlashCategory=2-Shot (Generalist VLMs)2025.10 | — | — | 21.87 | — | |
| Gemini-1.5-ProCategory=2-Shot (Generalist VLMs)2025.10 | — | — | 31.69 | — | |
| GPT 5.42026.07 | — | — | — | 78.3 | |
| GPT-4oCategory=2-Shot (Generalist VLMs)2025.10 | — | — | 26.52 | — | |
| GPT-o3Category=2-Shot (Generalist VLMs)2025.10 | — | — | 33.3 | — | |
| Grounded 3D-LLM3D Generalist=true2024.11 | — | 72.7 | — | — | |
| Grounded 3D-LLMModel category=Generalists, 3D geometric inputs=Yes2026.03 | — | 72.7 | — | — | |
| LL3DA3D Generalist=true2024.11 | — | 76.8 | — | — | |
| LL3DAModel category=Generalists, 3D geometric inputs=Yes2026.03 | — | 76.8 | — | — | |
| LL3DA2026.04 | — | — | 76.8 | — | |
| LLaVA-3DCategory=Fully Supervised (In-domain training)2025.10 | — | — | 103.1 | — | |
| LLaVA-Video-7B3D Generalist=false, zero-shot=true2024.11 | — | 88.7 | — | — | |
| MiMo Embodied-7B2026.07 | — | — | — | 83.7 | |
| MSSRCategory=2-Shot (Generalist VLMs)2025.10 | — | — | 32.49 | — | |
| Oryx-34B3D Generalist=false2024.11 | — | 72.3 | — | — | |
| Pelican VL-7B2026.07 | — | — | — | 65.6 | |
| Qwen2.5-VL-72BCategory=2-Shot (Generalist VLMs)2025.10 | — | — | 20.81 | — | |
| Qwen2.5-VL-7BCategory=2-Shot (Generalist VLMs)2025.10 | — | — | 18.32 | — | |
| RoboBrain 2.0-7B2026.07 | — | — | — | 59.7 | |
| RoboBrain 2.5-8B2026.07 | — | — | — | 63.1 | |
| RynnBrain 8B2026.07 | — | — | — | 60.2 | |
| VeBrain 7B2026.07 | — | — | — | 101.5 | |
| Video-3D-LLMCategory=Fully Supervised (In-domain training)2025.10 | — | — | 102.1 | — | |
| Vlaser 8B2026.07 | — | — | — | 55.5 |