Perception on MME-RealWorld Lite
67Overall ScorePFlowNet
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| PFlowNetBase Model=Qwen3-VL-8B2026.05 | 67 | 95.6 | 69.3 | 90 | 53.6 | 58.2 | |
| TTSPBackbone=Qwen3-VL-8B-Instruct2026.04 | 59 | 90.8 | 65.3 | 94 | 41.1 | 40 | |
| RTWIBackbone=Qwen3-VL-8B-Instruct2026.04 | 57.8 | 90.4 | 62.7 | 92 | 39.8 | 38.9 | |
| TTSPBackbone=Qwen3-VL-4B-Instruct2026.04 | 57.7 | 93.2 | 56 | 91 | 39.8 | 40 | |
| DeepconfBackbone=Qwen3-VL-8B-Instruct2026.04 | 57.2 | 90 | 61.3 | 90 | 39.5 | 38.9 | |
| Self-Cer.Backbone=Qwen3-VL-8B-Instruct2026.04 | 57.1 | 90 | 60 | 92 | 39.2 | 38.6 | |
| CISCBackbone=Qwen3-VL-8B-Instruct2026.04 | 56.8 | 90 | 60 | 89 | 39.2 | 38.6 | |
| TRTBackbone=Qwen3-VL-8B-Instruct2026.04 | 56.8 | 90 | 60 | 89 | 39.2 | 38.6 | |
| Self-Ref.Backbone=Qwen3-VL-8B-Instruct2026.04 | 56.7 | 90.4 | 60 | 89 | 38.9 | 38.3 | |
| RTWIBackbone=Qwen3-VL-4B-Instruct2026.04 | 56.6 | 92.4 | 54.7 | 89 | 39.2 | 38.3 | |
| SCBackbone=Qwen3-VL-8B-Instruct2026.04 | 56.6 | 90 | 58 | 89 | 39.2 | 38.6 | |
| DeepconfBackbone=Qwen3-VL-4B-Instruct2026.04 | 55.9 | 91.6 | 53.3 | 89 | 38.9 | 37.4 | |
| CISCBackbone=Qwen3-VL-4B-Instruct2026.04 | 55.7 | 91.2 | 53.3 | 89 | 38.6 | 37.4 | |
| SCBackbone=Qwen3-VL-4B-Instruct2026.04 | 55.5 | 91.2 | 53.3 | 89 | 38.6 | 36.9 | |
| TRTBackbone=Qwen3-VL-4B-Instruct2026.04 | 55.5 | 90.8 | 53.3 | 89 | 38.6 | 36.9 | |
| Self-Cer.Backbone=Qwen3-VL-4B-Instruct2026.04 | 55.5 | 90.8 | 54 | 89 | 38.6 | 36.9 | |
| DeepEyes2026.04 | 55.4 | 90 | 52.7 | 89 | 43.3 | 33.4 | |
| Self-Ref.Backbone=Qwen3-VL-4B-Instruct2026.04 | 55.4 | 91.2 | 53.3 | 89 | 38.2 | 36.9 | |
| TreeVGR-7BParameters=7B2025.07 | 54.9 | 87.6 | 50.7 | 83 | 47 | 43.4 | |
| TreeVGR2026.05 | 54.9 | 87.6 | 50.7 | 83 | 47 | 43.4 | |
| Thyme2026.05 | 54.4 | 90.4 | 56.7 | 86 | 46.3 | 38.5 | |
| BaseBackbone=Qwen3-VL-4B-Instruct2026.04 | 53.4 | 89.6 | 50.7 | 85 | 37.3 | 34.3 | |
| DeepEyes-7BParameters=7B2025.07 | 53.2 | 90 | 52.7 | 89 | 43.3 | 33.4 | |
| DeepEyes2026.05 | 53.2 | 90 | 52.7 | 89 | 43.3 | 33.4 | |
| Pixel-Reasoner2026.04 | 53.1 | 89.6 | 52 | 86 | 38.9 | 30.9 | |
| BaseBackbone=Qwen3-VL-8B-Instruct2026.04 | 52.9 | 89.2 | 54.7 | 86 | 34.8 | 33.1 | |
| DeepEyesV22026.05 | 52.4 | 85.6 | 49.3 | 89 | 45.8 | 33.4 | |
| InternVL3-78BParameters=78B2025.07 | 52.3 | 87.6 | 54.7 | 77 | 42.6 | 36.6 | |
| InternVL3-78BBackbone Parameters=78B2026.05 | 52.3 | 87.6 | 54.7 | 77 | 42.6 | 36.6 | |
| Qwen3-VL-32BBackbone Parameters=32B2026.05 | 52 | 91.6 | 47.3 | 96 | 36.1 | 42.9 | |
| InternVL3-38BBackbone Parameters=38B2026.05 | 51 | 85.6 | 56 | 71 | 42.6 | 40 | |
| Pixel-Reasoner-7BParameters=7B2025.07 | 49.7 | 89.6 | 52 | 86 | 38.9 | 30.9 | |
| Pixel-Reasoner2026.05 | 49.7 | 89.6 | 52 | 86 | 38.9 | 30.9 | |
| LLaVA-OneVision-72BParameters=72B2025.07 | 48.7 | 79.2 | 50.7 | 67 | 37.9 | 40 | |
| LLaVA-OV-72BBackbone Parameters=72B2026.05 | 48.7 | 79.2 | 50.7 | 67 | 37.9 | 40 | |
| Qwen3-VL-8BBackbone Parameters=8B2026.05 | 48.6 | 92.8 | 57.3 | 87 | 36.4 | 31.4 | |
| InternVL3-8BParameters=8B2025.07 | 47.9 | 83.6 | 49.3 | 75 | 34.5 | 36.9 | |
| InternVL3-8BBackbone Parameters=8B2026.05 | 47.9 | 83.6 | 49.3 | 75 | 34.5 | 36.9 | |
| Qwen3-VL-4BBackbone Parameters=4B2026.05 | 47.1 | 90.8 | 44.7 | 87 | 34.8 | 32.6 | |
| Qwen2.5-VL-32BBackbone Parameters=32B2026.05 | 45.6 | 87.2 | 40.7 | 83 | 29.5 | 40.7 | |
| Qwen2.5-VL-72BParameters=72B2025.07 | 43.7 | 90.8 | 34 | 87 | 27.9 | 30.6 | |
| LLaVA-OneVision-7BParameters=7B2025.07 | 43.7 | 80 | 40 | 56 | 31.7 | 39.4 | |
| LLaVA-OV-7BBackbone Parameters=7B2026.05 | 43.7 | 80 | 40 | 56 | 31.7 | 39.4 | |
| Qwen2.5-VL-72BBackbone Parameters=72B2026.05 | 43.7 | 90.8 | 34 | 87 | 27.9 | 30.6 | |
| Qwen2.5-VL-7BParameters=7B2025.07 | 42.3 | 87.6 | 32.7 | 83 | 27.3 | 30 | |
| Qwen2.5-VL-7BBackbone Parameters=7B2026.05 | 42.3 | 87.6 | 32.7 | 83 | 27.3 | 30 |