List your company

Knowledge-intensive image editing on UniREditBench (test)

81.01Real World ScoreGPT-4o
50.257258.241166.22574.2089Feb 2, 2026

Evaluation results

MethodLinks
GPT-4omodel_source=closed-source2026.0281.0162.0773.39
UniReasonreasoning_protocol=w/ Reasoning, reasoning_type=interleaved reasoning and refinement2026.0274.8265.370.06
Qwen-Image-Editreasoning_protocol=w/o Reasoning2026.0270.9541.9256.52
Seedream 4.0model_source=closed-source2026.0266.2245.3855.77
BAGELreasoning_protocol=w/ Reasoning, reasoning_type=textual reasoning only2026.0256.845.150.96
OmniGen2reasoning_protocol=w/o Reasoning2026.0253.6933.1443.41
Lumina-DiMOOreasoning_protocol=w/o Reasoning2026.0251.4445.6148.54