Visual content generation has advanced from single-image to multi-image workflows, yet existing agents remain largely plan-driven and lack systematic reflection mechanisms to correct mid-trajectory visual errors. To address this limitation, we propose VisionCreator-R1, a native visual generation agent with explicit reflection, together with a Reflection-Plan Co-Optimization (RPCO) training methodology. Through extensive experiments and trajectory-level analysis, we uncover reflection-plan optimization asymmetry in reinforcement learning (RL): planning can be reliably optimized via plan rewards, while reflection learning is hindered by noisy credit assignment. Guided by this insight, our RPCO first trains on the self-constructed VCR-SFT dataset with reflection-strong single-image trajectories and planning-strong multi-image trajectories, then co-optimization on VCR-RL dataset via RL. This yields our unified VisionCreator-R1 agent, which consistently outperforms Gemini2.5Pro on existing benchmarks and our VCR-bench covering single-image and multi-image tasks.
Cite
@article{arxiv.2603.08812,
title = {VisionCreator-R1: A Reflection-Enhanced Native Visual-Generation Agentic Model},
author = {Jinxiang Lai and Wenzhe Zhao and Zexin Lu and Hualei Zhang and Qinyu Yang and Rongwei Quan and Zhimin Li and Shuai Shao and Song Guo and Qinglin Lu},
journal= {arXiv preprint arXiv:2603.08812},
year = {2026}
}