Visual Question Answering on RWQA
80.8AccuracyGPT-5
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5Access=API call only2026.03 | 80.8 | |
| VISEModel Scale=32B2026.06 | 79.51 | |
| VisionZero-ChartModel Scale=32B2026.06 | 79.34 | |
| VisionZero-CLEVRModel Scale=32B2026.06 | 79.29 | |
| iReasonerModel Scale=32B2026.06 | 79.05 | |
| EvoLMMModel Scale=32B2026.06 | 78.98 | |
| VisionZero-RWModel Scale=32B2026.06 | 78.31 | |
| VisPlayModel Scale=32B2026.06 | 78.24 | |
| BaseModel Scale=32B2026.06 | 78.16 | |
| Molmo2-8BAccess=Fully open2026.03 | 77.6 | |
| Gemini 2.5 ProAccess=API call only2026.03 | 77.4 | |
| MolmoPoint-8BAccess=MolmoPoint2026.03 | 77.4 | |
| GPT-5 miniAccess=API call only2026.03 | 77 | |
| Eagle2.5-8BAccess=Open weights only2026.03 | 76.7 | |
| Molmo2-4BAccess=Fully open2026.03 | 75.4 | |
| PLM-8BAccess=Fully open2026.03 | 75 | |
| Qwen3-VL-8BTraining=Staged2026.05 | 74.51 | |
| Gemini 2.5 FlashAccess=API call only2026.03 | 74.5 | |
| MolmoPoint-8B-O-7BAccess=Fully open2026.03 | 73.6 | |
| Keye-VL-1.5-8BAccess=Open weights only2026.03 | 73.5 | |
| PLM-3BAccess=Fully open2026.03 | 72.4 | |
| MiniCPM-V-4.5-8BAccess=Open weights only2026.03 | 72.1 | |
| Qwen3-VL-8BAccess=Open weights only2026.03 | 71.5 | |
| OneThinker-8BTraining=Base2026.05 | 71.5 | |
| Qwen3-VL-4BAccess=Open weights only2026.03 | 70.9 | |
| Qwen3-VL-8BTraining=Base2026.05 | 70.85 | |
| GLM-4.1V-9BAccess=Open weights only2026.03 | 70.7 | |
| Qwen2.5-VL-7BTraining=Staged2026.05 | 70.46 | |
| VISEModel Scale=4B2026.06 | 70.46 | |
| VISEModel Scale=8B2026.06 | 70.03 | |
| VisionZero-ChartModel Scale=8B2026.06 | 69.95 | |
| VisionZero-CLEVRModel Scale=8B2026.06 | 69.88 | |
| iReasonerModel Scale=8B2026.06 | 69.81 | |
| GThinker-7BTraining=Base2026.05 | 69.8 | |
| VisionZero-RWModel Scale=4B2026.06 | 69.74 | |
| EvoLMMModel Scale=8B2026.06 | 69.74 | |
| WeThink-7BTraining=Base2026.05 | 69.54 | |
| OpenVLThinker-7BTraining=Base2026.05 | 69.41 | |
| VisPlayModel Scale=8B2026.06 | 69.41 | |
| VisionZero-RWModel Scale=8B2026.06 | 69.33 | |
| BaseModel Scale=8B2026.06 | 69.28 | |
| iReasonerModel Scale=4B2026.06 | 68.92 | |
| VisPlayModel Scale=4B2026.06 | 68.67 | |
| EvoLMMModel Scale=4B2026.06 | 68.65 | |
| VisionZero-CLEVRModel Scale=4B2026.06 | 68.43 | |
| VisionZero-ChartModel Scale=4B2026.06 | 67.91 | |
| BaseModel Scale=4B2026.06 | 67.82 | |
| InternVL3.5-8BAccess=Open weights only2026.03 | 67.5 | |
| Qwen2.5-VL-7BTraining=Base2026.05 | 67.45 | |
| InternVL3.5-4BAccess=Open weights only2026.03 | 66.3 | |
| VISEModel Scale=2B2026.06 | 64.58 | |
| MMR1-7BTraining=Base2026.05 | 64.31 | |
| VisionZero-RWModel Scale=2B2026.06 | 64.31 | |
| iReasonerModel Scale=2B2026.06 | 63.94 | |
| EvoLMMModel Scale=2B2026.06 | 63.78 | |
| VisionZero-CLEVRModel Scale=2B2026.06 | 63.75 | |
| VisionZero-ChartModel Scale=2B2026.06 | 63.64 | |
| VisPlayModel Scale=2B2026.06 | 63.62 | |
| InstructVLA-Generalistnumber of parameters=1.5B, robot state=true2025.07 | 63.5 | |
| BaseModel Scale=2B2026.06 | 63.41 | |
| R1-OneVision-RL-7BTraining=Base2026.05 | 63.4 | |
| Eagle2number of parameters=1.5B2025.07 | 63.1 | |
| InstructVLA-Generalistnumber of parameters=1.5B2025.07 | 63.1 | |
| Qwen2-VLnumber of parameters=1.5B2025.07 | 62.9 | |
| Claude Sonnet 4.5Access=API call only2026.03 | 61.1 | |
| Bunnynumber of parameters=8B2025.07 | 60.4 | |
| VGR-7BDownSample=2×2 | 4×4, #Vtoken=864, LLM=Vicuna-7B2025.06 | 59.8 | |
| VGR-7BDownSample=2×2 | 2×2, #Vtoken=3024, LLM=Vicuna-7B2025.06 | 59.5 | |
| LLaVA-NeXT-7BDownSample=No, #Vtoken=2880, LLM=Vicuna-7B2025.06 | 57.8 | |
| ChatVLAnumber of parameters=1.5B2025.07 | 57 | |
| LLaVA-NeXT-7B†DownSample=2×2 | 4×4, #Vtoken=864, LLM=Vicuna-7B2025.06 | 56.8 | |
| Magmanumber of parameters=8B2025.07 | 56.5 | |
| PaliGemmanumber of parameters=2B2025.07 | 55.2 | |
| LLaVA-v1.5-7BDownSample=No, #Vtoken=576, LLM=Vicuna-7B2025.06 | 54.8 | |
| DeepSeek-VL-7BDownSample=Conv2D, #Vtoken=576, LLM=DeepSeek-7B2025.06 | 54.2 | |
| Qwen-VL-Chat-7BDownSample=Cross-Attn, #Vtoken=1024, LLM=Qwen-7B2025.06 | 49.3 | |
| OpenVLAnumber of parameters=7B, fine-tuned=true2025.07 | 47.2 | |
| ECoTnumber of parameters=7B2025.07 | 29.8 | |
| OpenVLAnumber of parameters=7B2025.07 | 0 |