KRIS-Bench: Benchmarking Next-Level
Intelligent Image Editing Models

Yongliang Wu1,4, Zonghui Li1, Xinting Hu2†, Xinyu Ye3, Xianfang Zeng4‡, Gang Yu4, Wenbo Zhu5, Bernt Schiele2, Ming-Hsuan Yang6, Xu Yang1
1Southeast University, 2Max Planck Institute for Informatics, 3Shanghai Jiao Tong University 4StepFun, 5University of California, Berkeley, 6University of California, Merced
†Corresponding author, ‡Project leader
Paper Code

🤗

 KRIS-Bench

🤗

 Results Gallery
  • [2026-05-06] 🔥 We update the evaluation results of Step Image Edit 2.
  • [2026-05-06] 🔥 We update the evaluation results of JoyAI-Image-Edit.
  • [2026-02-01] 🔥 We update the evaluation results of Step1x-Edit v1.2.
  • [2026-02-01] 🔥 We update the evaluation results of Qwen-Image-Edit-2511.
  • [2025-09-19] 🎉 KRIS-Bench is accepted by NeurIPS 2025.
  • [2025-07-28] 🔥 We update the evaluation results of Flux.1 Kontext [Pro] and [Max].
  • [2025-07-24] 🔥 We update the evaluation results of Step 3o vision
  • [2025-07-15] 🔥 We update the evaluation results of Step1X-Edit v1.1
  • [2025-07-01] 🔥 We update the evaluation results of Flux.1 Kontext [Dev].
  • [2025-06-26] 🔥 We update the evaluation results of HiDream-E1.
  • [2025-06-25] 🔥 We update the evaluation results of OmniGen2.
  • [2025-06-19] 🔥 We update the evaluation results of ByteMorph.
  • [2025-06-18] 🔥 We update the official evaluation results from BAGEL team, please refer to this link.
  • [2025-06-11] 🔥 We update the evaluation results of UniWorld-V1.

Abstract

Recent advances in multi-modal generative models have enabled significant progress in instruction-based image editing. However, while these models produce visually plausible outputs, their capacity for knowledge-based reasoning editing tasks remains under-explored. In this paper, We introduce KRIS-Bench (Knowledge-based Reasoning in Image-editing Systems Benchmark), a diagnostic benchmark designed to assess models through a cognitively informed lens. Drawing from educational theory, KRIS-Bench categorizes editing tasks across three foundational knowledge types: Factual, Conceptual, and Procedural. Based on this taxonomy, we design 22 representative tasks spanning 7 reasoning dimensions and release 1,267 high-quality annotated editing instances. To support fine-grained evaluation, we propose a comprehensive protocol that incorporates a novel Knowledge Plausibility metric, enhanced by knowledge hints and calibrated through human studies. Empirical results on nine state-of-the-art models reveal significant gaps in reasoning performance, highlighting the need for knowledge-centric benchmarks to advance the development of intelligent image editing systems.

KRIS-Bench Overview

Leaderboard

Model Factual Knowledge Conceptual Knowledge Procedural Knowledge Overall Score
Attribute Perception Spatial Perception Temporal Perception Average Score Social Science Natural Science Average Score Logical Reasoning Instruction Decomposition Average Score
GPT-4o

OpenAI

 		83.17 79.08 68.25 79.80 85.50 80.06 81.37 71.56 85.08 78.32 80.09
Step Image Edit 2

StepFun

 		81.49 78.25 0 67.34 74.04 69.28 70.44 68.55 45.45 56.99 66.16
JoyAI-Image-Edit

Jingdong

 		78.09 80.33 0 65.72 66.56 64.6 65.09 68.02 46.33 57.17 63.44
Qwen-Image-Edit-2511

Alibaba

 		76.3 85.08 0 65.7 69.15 61.97 63.67 61.54 45.89 53.71 62.04
Step1x-Edit v1.2

StepFun

 		76.33 70.42 0 62.44 69.56 64.5 65.72 60 40.83 50.42 60.93
Gemini 2.0

Google

 		66.33 63.33 63.92 65.26 68.19 56.94 59.65 54.13 71.67 62.90 62.41
Step 3o vision

StepFun

 		69.67 61.08 63.25 66.70 66.88 60.88 62.32 49.06 54.92 51.99 61.43
Doubao

ByteDance

 		70.92 59.17 40.58 63.30 65.50 61.19 62.23 47.75 60.58 54.17 60.70
BAGEL-Think

ByteDance

 		67.42 68.33 58.67 66.18 63.55 61.40 61.92 48.12 50.22 49.02 60.18
BAGEL

ByteDance

 		64.27 62.42 42.45 60.26 55.40 56.01 55.86 52.54 50.56 51.69 56.21
FLUX.1 Kontext [Max]

Black Forest Labs

 		71.25 69.17 0.00 59.04 60.88 56.06 57.22 50.38 40.83 45.60 55.12
FLUX.1 Kontext [Pro]

Black Forest Labs

 		69.42 70.17 0.00 58.14 55.44 54.94 55.06 50.12 43.25 46.69 54.17
Step1X-Edit v1.1

StepFun

 		64.17 61.75 0.00 53.05 52.06 55.06 54.34 52.56 36.75 44.66 51.59
UniWorld-V1

PKU

 		58.17 54.50 63.00 47.71 47.50 43.94 44.80 42.00 53.83 47.92 50.27
OmniGen2

BAAI

 		59.92 52.25 54.75 57.36 47.56 43.12 44.20 32.50 63.08 47.79 49.71
FLUX.1 Kontext [Dev]

Black Forest Labs

 		64.83 60.92 0.00 53.28 48.94 50.81 50.36 46.06 39.00 42.53 49.54
ByteMorph

ByteDance

 		61.17 62.00 0.00 51.27 45.50 47.38 46.92 32.00 31.33 31.67 44.85
HiDream-E1

HiDream.ai

 		52.75 49.42 0.00 43.31 52.56 49.25 50.05 45.19 30.08 37.64 44.72
Step1X-Edit

StepFun

 		55.50 51.75 0.00 45.52 44.69 49.06 48.01 40.88 22.75 31.82 43.29
Emu2

BAAI

 		51.50 48.83 22.17 45.40 34.69 38.44 37.54 24.81 45.00 34.91 39.70
AnyEdit

ZJU

 		47.67 45.17 0.00 39.26 38.56 42.94 41.88 36.56 26.92 31.74 38.55
MagicBrush

OSU

 		53.92 39.58 0.00 41.84 42.94 38.06 39.24 30.00 23.08 26.54 37.15
OmniGen

BAAI

 		37.92 28.25 21.83 33.11 30.63 27.19 28.02 11.94 35.83 23.89 28.85
InstructPix2Pix

UCB

 		30.33 21.33 0.00 23.33 22.56 26.56 25.59 19.81 14.75 17.28 22.82

Note: If you would like to submit your results, please contact us at yongliang0223@gmail.com.

Benchmark Examples

KRIS-Bench Method

BibTeX

@article{wu2026kris,
        title={Kris-bench: Benchmarking next-level intelligent image editing models},
        author={Wu, Yongliang and Li, Zonghui and Hu, Xinting and Ye, Xinyu and Zeng, Xianfang and Yu, Gang and Zhu, Wenbo and Schiele, Bernt and Yang, Ming-Hsuan and Yang, Xu},
        journal={Advances in Neural Information Processing Systems},
        volume={38},
        year={2026}
      }