Text-to-Image on GenExam Math (Overall)
68.2Strict AccuracyDraw2Think
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Draw2ThinkModel Category=Closed-source Models, Base Model / Release Information=Gemini 3 Flash Preview | Dec. 20252026.05 | 68.2 | 90.5 | |
| Nano Banana 2Model Category=Closed-source Models, Base Model / Release Information=Gemini 3.1 Flash Image | Feb. 20262026.05 | 56.3 | 87.8 | |
| Nano Banana ProModel Category=Closed-source Models, Base Model / Release Information=Gemini 3 Pro Image | Nov. 20252026.05 | 55.6 | 86.3 | |
| GPT-Image-2Model Category=Closed-source Models, Base Model / Release Information=OpenAI native T2I | Apr. 20262026.05 | 50.3 | 85.2 | |
| Seedream 5.0Model Category=Closed-source Models, Base Model / Release Information=ByteDance native T2I | Feb. 20262026.05 | 47 | 82.9 | |
| GPT-Image-1.5Model Category=Closed-source Models, Base Model / Release Information=OpenAI native T2I | Dec. 20252026.05 | 26.5 | 65.8 | |
| FaireModel Category=Open-source Code-rendered RL, Base Model / Release Information=Qwen3-VL-8B + GRPO2026.05 | 9.3 | 52.3 | |
| FLUX.2 maxModel Category=Closed-source Models, Base Model / Release Information=Black Forest Labs T2I | Dec. 20252026.05 | 6.6 | 49.1 | |
| FLUX.2 devModel Category=Open-source T2I Models2026.05 | 2.6 | 31.6 | |
| Qwen-Image-2512Model Category=Open-source T2I Models2026.05 | 0 | 27.9 | |
| HunyuanImage-3.0Model Category=Open-source T2I Models2026.05 | 0 | 17 | |
| BLIP3o-NEXT-GRPOModel Category=Open-source Unified MLLMs2026.05 | 0 | 15.5 | |
| BAGELModel Category=Open-source Unified MLLMs2026.05 | 0 | 14.7 | |
| Janus-ProModel Category=Open-source Unified MLLMs2026.05 | 0 | 13.7 |