Skywork-R1V4: Toward Agentic Multimodal Intelligence through Interleaved Thinking with Images and DeepResearch
Research paper by Yifan Zhang and collaborators. Published 2025-12-02. Topics: multimodal agentic systems, image manipulation, web search, reinforcement learning, multimodal planning.
Resource information
- Authors
- Yifan Zhang, Liang Hu, Haofeng Sun, Peiyu Wang, Yichen Wei, Shukang Yin, Jiangbo Pei, Wei Shen, Peng Xia, Yi Peng, Tianyidan Xie, Eric Li, Yang Liu, Xuchen Song, Yahui Zhou
- Research area
- Computer vision
- Application domain
- Agent systems
- AI methods
- Reinforcement learning
- Research topics
- cs.CV
- arXiv record date
- 2025-12-08
- Publication date
- 2025-12-02
- Publication
- arXiv
- Upstream last modified
- Dec 9, 2025
- Last record update
- Nov 3, 2026
Abstract
Despite recent progress in multimodal agentic systems, existing approaches often treat image manipulation and web search as disjoint capabilities, rely heavily on costly reinforcement learning, and lack planning grounded in real tool-execution traces. To address these limitations, we present Skywork-R1V4, a 30B (A3B) parameter multimodal agentic model that unifies multimodal planning, active image manipulation ("thinking with images"), deep multimodal search, and, most critically, interleaved reasoning that dynamically alternates between visual operations and external knowledge retrieval. Trained solely via supervised fine-tuning on fewer than 30,000 high-quality, planning-execution-consistent trajectories and validated through stepwise consistency filtering, Skywork-R1V4 achieves state-of-the-art results across perception and multimodal search benchmarks: it scores 66.1 on MMSearch and 67.2 on FVQA, surpassing Gemini 2.5 Flash on all 11 metrics. Skywork-R1V4 exhibits emergent long-horizon reasoning at inference time, successfully orchestrating more than 10 tool calls to solve complex, multi-step tasks. Our results demonstrate that sophisticated agentic multimodal intelligence can be achieved through carefully curated supervised learning alone, without any reliance on reinforcement learning.