Text-to-Image Generation on GenEval++
90Color AccuracyGPT4o
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| GPT4oType=Unified2026.01 | 90 | 67.5 | 72.5 | 62.5 | 60 | 80 | 85 | 73.9 | |
| GPT-4o2026.05 | 90 | 67.5 | 72.5 | 62.5 | 60 | 80 | 85 | 73.9 | |
| CLVRModel Size=4B2026.05 | 90 | 45 | 45 | 65 | 60 | 72.5 | 53.8 | 61.6 | |
| GPT-Image-1Model Type=Closed-source2026.05 | 90 | 67.5 | 72.5 | 62.5 | 60 | 80 | 85 | 73.9 | |
| Qwen-ImageType=Diffusion2026.01 | 87.5 | 72.5 | 72.5 | 60 | 47.5 | 72.5 | 55 | 66.8 | |
| CLVRModel Size=9B2026.05 | 87.5 | 62.5 | 60 | 75 | 57.5 | 70 | 70 | 68.9 | |
| Qwen-ImageModel Type=Open-source2026.05 | 87.5 | 72.5 | 72.5 | 60 | 47.5 | 72.5 | 55 | 66.8 | |
| GenClawModel Type=Agentic generation2026.05 | 87.5 | 95 | 82.5 | 85 | 92.5 | 82.5 | 90 | 87.8 | |
| GPT-Image-1.5Model Type=Closed-source2026.05 | 85 | 95 | 77.5 | 50 | 60 | 77.5 | 80 | 75 | |
| Gemini-3.1 Flash-ImageModel Type=Closed-source2026.05 | 85 | 85 | 80 | 75 | 67.5 | 72.5 | 77.5 | 77.5 | |
| Nano-BananaModel Type=Closed-source2026.05 | 80 | 70 | 68 | 53 | 68 | 78 | 80 | 71 | |
| Gemini-3.0 Pro-ImageModel Type=Agentic generation2026.05 | 80 | 90 | 75 | 70 | 72.5 | 72.5 | 85 | 76.1 | |
| GenAgent (w/ Qwen-Image)Type=GenAgent (Ours), Variant=w/ Qwen-Image2026.01 | 77.5 | 77.5 | 65 | 80 | 60 | 72.5 | 75 | 72.5 | |
| GenAgentModel Type=Agentic generation2026.05 | 77.5 | 77.5 | 65 | 80 | 60 | 72.5 | 75 | 72.5 | |
| Mind-BrushModel Type=Agentic generation2026.05 | 77.5 | 70 | 77.5 | 75 | 85 | 77.5 | 85 | 78.2 | |
| GenAgent (+RL)Type=GenAgent (Ours), Variant=+RL2026.01 | 75 | 67.5 | 37.5 | 52.5 | 45 | 62.5 | 52.5 | 56.1 | |
| FLUX.2 [Klein]Model Size=4B2026.05 | 75 | 42.5 | 22.5 | 30 | 5 | 45 | 42.5 | 37.5 | |
| FLUX.2 [Klein]Model Size=9B2026.05 | 75 | 45 | 25 | 0 | 5 | 27.5 | 37.5 | 30.7 | |
| Uni-CoT2026.05 | 70 | 67.5 | 65.5 | 62.5 | 60 | 62.5 | 57.5 | 63.5 | |
| T2I-R1Method Paradigm=Unified MLLM w/ CoT2025.12 | 68 | 33 | 20 | 35 | 8 | 25 | 30 | 31 | |
| T2I-R1Type=Unified2026.01 | 67.5 | 32.5 | 20 | 35 | 7.5 | 25 | 30 | 31.1 | |
| T2I-R12025.12 | 67.5 | 32.5 | 20 | 35 | 7.5 | 25 | 30 | 31.1 | |
| T2I-R12026.05 | 67.5 | 32.5 | 20 | 35 | 7.5 | 25 | 30 | 31.1 | |
| T2I-R1Model Type=Open-source2026.05 | 67.5 | 32.5 | 20 | 35 | 7.5 | 25 | 30 | 31.1 | |
| BLIP3-oModel Type=Open-source2026.05 | 67.5 | 32.5 | 20 | 35 | 7.5 | 25 | 30 | 31.1 | |
| OmniGen2Model Type=Open-source2026.05 | 67.5 | 32.5 | 20 | 35 | 7.5 | 25 | 30 | 31.1 | |
| GenAgent (Ours w/o tool)Type=Decoupled, Variant=w/o tool2026.01 | 65 | 57.5 | 35 | 45 | 35 | 60 | 40 | 48.2 | |
| SDv3-medium2025.12 | 55 | 50 | 12.5 | 35 | 17.5 | 15 | 22.5 | 29.6 | |
| OmniGen22025.12 | 55 | 42.5 | 20 | 27.5 | 12.5 | 25 | 45 | 32.5 | |
| NPC2025.12 | 55 | 67.5 | 35 | 52.5 | 55 | 72.5 | 62.5 | 57.1 | |
| SD3-mediumModel Type=Only Generation2026.03 | 55 | 50 | 12.5 | 35 | 17.5 | 15 | 22.5 | 29.6 | |
| OmniGen2Model Type=Generation & Understanding2026.03 | 55 | 42.5 | 20 | 27.5 | 12.5 | 25 | 45 | 32.5 | |
| OmniGen22026.05 | 55 | 42.5 | 20 | 27.5 | 12.5 | 25 | 45 | 32.5 | |
| PromptEnhancerType=Decoupled2026.01 | 50 | 62.5 | 22.5 | 37.5 | 12.5 | 45 | 37.5 | 38.2 | |
| GenAgent (+SFT)Type=GenAgent (Ours), Variant=+SFT2026.01 | 50 | 55 | 55 | 52.5 | 25 | 60 | 57.5 | 50.7 | |
| PromptEnhancerModel Type=Agentic generation2026.05 | 50 | 62.5 | 22.5 | 37.5 | 12.5 | 45 | 37.5 | 38.2 | |
| Janus Pro 7BType=Unified2026.01 | 45 | 30 | 12.5 | 30 | 7.5 | 35 | 12.5 | 24.6 | |
| Janus-Pro2025.12 | 45 | 30 | 12.5 | 30 | 7.5 | 35 | 12.5 | 24.6 | |
| Janus-Pro-7BMethod Paradigm=Unified MLLM2025.12 | 45 | 30 | 13 | 30 | 8 | 35 | 13 | 25 | |
| DraCoMethod Paradigm=Unified MLLM w/ CoT2025.12 | 45 | 53 | 28 | 40 | 28 | 53 | 38 | 40 | |
| Janus-ProModel Type=Generation & Understanding2026.03 | 45 | 30 | 12.5 | 30 | 7.5 | 35 | 12.5 | 24.6 | |
| Janus-Pro2026.05 | 45 | 30 | 12.5 | 30 | 7.5 | 35 | 12.5 | 24.6 | |
| Janus Pro 7BModel Type=Open-source2026.05 | 45 | 30 | 12.5 | 30 | 7.5 | 35 | 12.5 | 24.6 | |
| DraCoModel Type=Open-source2026.05 | 45 | 53 | 28 | 40 | 28 | 53 | 38 | 40 | |
| FLUX.1-Kontext2025.12 | 42.5 | 50 | 20 | 25 | 30 | 40 | 32.5 | 34.3 | |
| FLUX.1-Kontext2026.05 | 42.5 | 50 | 20 | 25 | 30 | 40 | 32.5 | 34.3 | |
| FLUX.1-devType=Diffusion2026.01 | 40 | 60 | 25 | 25 | 7.5 | 40 | 30 | 32.5 | |
| ReflectionFlowType=Decoupled2026.01 | 40 | 62.5 | 27.5 | 27.5 | 20 | 42.5 | 32.5 | 36.1 | |
| SD-3.5 MModel Type=Open-source2026.05 | 40 | 60 | 25 | 25 | 7.5 | 40 | 30 | 32.5 | |
| FLUX.1-devModel Type=Open-source2026.05 | 40 | 60 | 25 | 25 | 7.5 | 40 | 30 | 32.5 | |
| FLUX.1-dev2025.12 | 35 | 62.5 | 15 | 27.5 | 20 | 37.5 | 22.5 | 31.4 | |
| FLUX.1-devMethod Paradigm=Generation Only2025.12 | 35 | 63 | 15 | 28 | 20 | 38 | 23 | 31 | |
| Bagel-ThinkMethod Paradigm=Unified MLLM w/ CoT2025.12 | 35 | 38 | 35 | 45 | 13 | 48 | 35 | 35 | |
| FLUX.1-devModel Type=Only Generation2026.03 | 35 | 62.5 | 15 | 27.5 | 20 | 37.5 | 22.5 | 31.4 | |
| FLUX.1-dev2026.05 | 35 | 62.5 | 15 | 27.5 | 20 | 37.5 | 22.5 | 31.4 | |
| BagelMethod Paradigm=Unified MLLM2025.12 | 33 | 60 | 25 | 33 | 25 | 48 | 38 | 37 | |
| BagelType=Unified2026.01 | 32.5 | 60 | 25 | 32.5 | 25 | 47.5 | 37.5 | 37.1 | |
| Bagel2025.12 | 32.5 | 60 | 25 | 32.5 | 25 | 47.5 | 37.5 | 37.1 | |
| BAGELModel Type=Generation & Understanding2026.03 | 32.5 | 60 | 25 | 32.5 | 25 | 47.5 | 37.5 | 37.1 | |
| BAGEL2026.05 | 32.5 | 60 | 25 | 32.5 | 25 | 47.5 | 37.5 | 37.1 | |
| BagelModel Type=Open-source2026.05 | 32.5 | 60 | 25 | 32.5 | 25 | 47.5 | 37.5 | 37.1 | |
| GenAgent (Base)Type=GenAgent (Ours), Variant=Base2026.01 | 30 | 60 | 32.5 | 30 | 32.5 | 30 | 35 | 35.7 | |
| GvUModel Type=Generation & Understanding2026.03 | 30 | 40 | 15 | 57.5 | 52.5 | 67.5 | 40 | 40.4 | |
| BLIP3-oModel size=8B2025.12 | 25 | 25 | 12.5 | 60 | 12.5 | 57.5 | 22.5 | 30.7 | |
| BLIP3-o 8BMethod Paradigm=Unified MLLM2025.12 | 25 | 25 | 13 | 60 | 13 | 58 | 23 | 31 | |
| BLIP3-o 8BModel Type=Generation & Understanding2026.03 | 25 | 25 | 12.5 | 60 | 12.5 | 57.5 | 22.5 | 30.7 | |
| XOmniModel Type=Generation & Understanding2026.03 | 22.5 | 50 | 2.5 | 32.5 | 15 | 47.5 | 27.5 | 28.2 | |
| BLIP3-o 4BMethod Paradigm=Unified MLLM2025.12 | 13 | 23 | 10 | 45 | 13 | 55 | 23 | 26 | |
| BLIP3-oModel size=4B2025.12 | 12.5 | 22.5 | 10 | 45 | 12.5 | 55 | 22.5 | 25.7 | |
| SDXL2025.12 | 5 | 37.5 | 0 | 0 | 0 | 0 | 0 | 6.1 | |
| SDXLMethod Paradigm=Generation Only2025.12 | 5 | 38 | 0 | 0 | 0 | 0 | 0 | 6 | |
| SD XLModel Type=Only Generation2026.03 | 5 | 37.5 | 0 | 0 | 0 | 0 | 0 | 6.1 | |
| SDv2.12025.12 | 0 | 32.5 | 2.5 | 0 | 0 | 2.5 | 7.5 | 6.4 | |
| SDv2.1Method Paradigm=Generation Only2025.12 | 0 | 33 | 3 | 0 | 0 | 3 | 8 | 6 | |
| SD v2.1Model Type=Only Generation2026.03 | 0 | 32.5 | 2.5 | 0 | 0 | 2.5 | 7.5 | 6.4 |