Spatial Reasoning Generation on GenEval (test)
90.3Mean ScoreSAGen
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| SAGenModel Category=MLLM Enhanced Generation Models, Chain-of-Thought=true, Strategy=SCoT2026.02 | 90.3 | 98.1 | 94.2 | 90 | 87.2 | 90.5 | 81.5 | |
| BAGELModel Category=MLLM Enhanced Generation Models, Chain-of-Thought=true2026.02 | 88 | 98 | 95 | 84 | 95 | 78 | 77 | |
| Qwen-ImageModel Category=MLLM Enhanced Generation Models2026.02 | 87 | 99 | 92 | 89 | 88 | 76 | 77 | |
| BLIP3o-8BModel Category=MLLM Enhanced Generation Models2026.02 | 84 | — | — | — | — | — | — | |
| GPT-ImageModel Category=Closed-Source Generation Models2026.02 | 84 | 99 | 92 | 85 | 92 | 75 | 61 | |
| HiDream-I1Model Category=Diffusion Models2026.02 | 83 | 100 | 98 | 79 | 91 | 60 | 72 | |
| SD-3.5-LargeModel Category=Diffusion Models2026.02 | 71 | 98 | 89 | 73 | 83 | 34 | 47 | |
| FLUX.1-devModel Category=Diffusion Models2026.02 | 66 | 98 | 81 | 74 | 79 | 22 | 45 | |
| SD-3.5-MediumModel Category=Diffusion Models2026.02 | 63 | 98 | 78 | 50 | 81 | 24 | 52 |