Real-world visual perception on MME-RealWorld CN
73.31AccuracyGemini-3.1-Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini-3.1-ProParam Size=-, Inference Mode=Closed-Source2026.05 | 73.31 | |
| GPT-5.4Param Size=-, Inference Mode=Closed-Source2026.05 | 70.93 | |
| Vision-OPD (Ours)Param Size=4B, Inference Mode=Single Forward Pass2026.05 | 70.76 | |
| Vision-OPD (Ours)Param Size=9B, Inference Mode=Single Forward Pass2026.05 | 70.46 | |
| Qwen3.5Param Size=397B, Inference Mode=Single Forward Pass2026.05 | 69.82 | |
| Qwen3-VL-InstructParam Size=235B, Inference Mode=Single Forward Pass2026.05 | 69.04 | |
| SenseNova-MARSParam Size=8B, Inference Mode=Agentic2026.05 | 68.9 | |
| GPT-5.2Param Size=-, Inference Mode=Closed-Source2026.05 | 68.8 | |
| ZwZParam Size=8B, Inference Mode=Single Forward Pass2026.05 | 68.09 | |
| Qwen3.5Param Size=9B, Inference Mode=Single Forward Pass2026.05 | 67.67 | |
| Kimi-K2.5Param Size=1T, Inference Mode=Single Forward Pass2026.05 | 65.71 | |
| GLM-4.6VParam Size=106B, Inference Mode=Single Forward Pass2026.05 | 65.62 | |
| DeepEyesV2Param Size=7B, Inference Mode=Agentic2026.05 | 65.07 | |
| Qwen3-VL-InstructParam Size=8B, Inference Mode=Single Forward Pass2026.05 | 64.61 | |
| ThymeParam Size=7B, Inference Mode=Agentic2026.05 | 64.59 | |
| DeepEyesParam Size=7B, Inference Mode=Agentic2026.05 | 64.09 | |
| Qwen3.5Param Size=4B, Inference Mode=Single Forward Pass2026.05 | 63.7 | |
| MiniCPM-V-4.5Param Size=9B, Inference Mode=Single Forward Pass2026.05 | 61.64 | |
| MiMo-VL-RLParam Size=7B, Inference Mode=Single Forward Pass2026.05 | 55.89 |