# EditJev: System One Routing for Image Editing Models

DevFeed: [EditJev: System One Routing for Image Editing Models](<https://devfeed.tech/articles/editjev-system-one-routing-for-image-editing-models-100028.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.32635>)

Author: Zhile Guan, Xu Jiang, Yule Duan, Gongbo Zhang, Shuyu Wang, Xiao Chen, Jian Zhang

Published: 2026-10-26T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Model Routing](<https://devfeed.tech/topics/model-routing.md>), [image editing](<https://devfeed.tech/topics/image-editing.md>), [vlm](<https://devfeed.tech/topics/vlm.md>)

Tags: [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [image-editing](<https://devfeed.tech/tags/image-editing.md>), [latency](<https://devfeed.tech/tags/latency.md>), [model-routing](<https://devfeed.tech/tags/model-routing.md>), [performance](<https://devfeed.tech/tags/performance.md>), [quality](<https://devfeed.tech/tags/quality.md>), [ranking](<https://devfeed.tech/tags/ranking.md>), [routing](<https://devfeed.tech/tags/routing.md>), [score](<https://devfeed.tech/tags/score.md>), [vision](<https://devfeed.tech/tags/vision.md>)

## AI overview

EditJev routes each image editing request to a suitable model, addressing variation in both model quality and inference speed. A lightweight preference head on a vision-language backbone predicts scores for candidate editors in one forward pass; it is trained on pairwise preferences from offline outputs. At inference, the scores are combined with model speed scores to balance quality and latency. On ImgEdit, aggregate quality rises from 4.540 to 4.592 while estimated end-to-end latency falls 60.8%; on GEdit-En, quality rises from 8.194 to 8.255 while latency falls 31.7%, compared with the strongest fixed editor. The supplied text reports these results but does not describe the candidate pool or evaluation setup in detail.

## Source excerpt

arXiv:2610.32635v1 Announce Type: cross Abstract: Instruction-based image editing models exhibit pronounced sample-specific performance variation: different models can produce substantially different results for the same source image and editing instruction, and the best-performing model often varies across requests. These models also differ substantially in inference latency, making fixed-model deployment inefficient. We therefore study image editing model routing, which dynamically selects an editor from a candidate pool given an input image and editing instruction. We introduce EditJev, a System One routing model that directly predicts sample-level preference scores over candidate image editing models in a single forward pass. A lightweight preference head on a vision-language backbone produces a fixed-dimensional score vector without autoregressive model-name generation or response parsing. To train EditJev, we construct pairwise preferences from offline candidate outputs and optimize a pairwise ranking objective to learn relative model suitability. At inference time, EditJev combines the predicted preference scores with model-level speed scores to balance editing quality and generation latency. Experiments on ImgEdit and GEdit-En show that EditJev improves aggregate quality over the strongest fixed editor, from 4.540 to 4.592 and from 8.194 to 8.255, respectively, while reducing estimated end-to-end latency by 60.8% and 31.7%. These results demonstrate the effectiveness of direct, structured model routing for efficient image editing.