Text-to-Image Generation on T2I-CoreBench
86.3Composition ScoreOmniVerifier-M1+GPT-Image-1.5
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| OmniVerifier-M1+GPT-Image-1.5Generative Model=GPT-Image-1.52026.05 | 86.3 | 76.9 | 80 | |
| Qwen3-VL 8B+GPT-Image-1.5Generative Model=GPT-Image-1.52026.05 | 85.7 | 75.2 | 78.7 | |
| GPT-Image-1.52026.05 | 85.5 | 74.6 | 78.2 | |
| OmniVerifier-M1+RePlanGenerative Model=RePlan2026.05 | 81.7 | 62.6 | 69 | |
| Qwen3-VL 8B+RePlanGenerative Model=RePlan2026.05 | 79.3 | 54.7 | 62.9 | |
| Qwen-Image2026.05 | 78 | 49.3 | 58.9 |