# Training Multi-Image Vision Agents via End2End Reinforcement Learning

arXiv: https://arxiv.org/abs/2512.08980 · HTML: https://mlsanity.com/paper/2512.08980/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 57 of its 69 references that are in the map.
- First posted on arXiv 5 Dec 2025.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)
- LLM agents and tool use / [Multimodal and embodied agents](https://mlsanity.com/agents/multimodal-and-embodied-agents/)

## Builds on (most cited first, 30 of 57)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [GPT-4o System Card](https://mlsanity.com/paper/2410.21276/) (2024)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [LLaVA-OneVision: Easy Visual Task Transfer](https://mlsanity.com/paper/2408.03326/) (2024)
- [Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning](https://mlsanity.com/paper/2503.09516/) (2025)
- [BLINK: Multimodal Large Language Models Can See but Not Perceive](https://mlsanity.com/paper/2404.12390/) (2024)
- [Video-R1: Reinforcing Video Reasoning in MLLMs](https://mlsanity.com/paper/2503.21776/) (2025)
- [V*: Guided Visual Search as a Core Mechanism in Multimodal LLMs](https://mlsanity.com/paper/2312.14135/) (2023)
- [RLHF-V: Towards Trustworthy MLLMs via Behavior Alignment from Fine-grained Correctional Human Feedback](https://mlsanity.com/paper/2312.00849/) (2023)
- [ReTool: Reinforcement Learning for Strategic Tool Use in LLMs](https://mlsanity.com/paper/2504.11536/) (2025)
- [MME-RealWorld: Could Your Multimodal LLM Challenge High-Resolution Real-World Scenarios that are Difficult for Humans?](https://mlsanity.com/paper/2408.13257/) (2024)
- [ToRA: A Tool-Integrated Reasoning Agent for Mathematical Problem Solving](https://mlsanity.com/paper/2309.17452/) (2023)
- [Pixel Reasoner: Incentivizing Pixel-Space Reasoning with Curiosity-Driven Reinforcement Learning](https://mlsanity.com/paper/2505.15966/) (2025)
- [Monte Carlo Tree Search Boosts Reasoning via Iterative Preference Learning](https://mlsanity.com/paper/2405.00451/) (2024)
- [Seg-Zero: Reasoning-Chain Guided Segmentation via Cognitive Reinforcement](https://mlsanity.com/paper/2503.06520/) (2025)
- [MANTIS: Interleaved Multi-Image Instruction Tuning](https://mlsanity.com/paper/2405.01483/) (2024)
- [The Landscape of Agentic Reinforcement Learning for LLMs: A Survey](https://mlsanity.com/paper/2509.02547/) (2025)
- [Thyme: Think Beyond Images](https://mlsanity.com/paper/2508.11630/) (2025)
- [R1-Zero's "Aha Moment" in Visual Reasoning on a 2B Non-SFT Model](https://mlsanity.com/paper/2503.05132/) (2025)
- [OpenThinkIMG: Learning to Think with Images via Visual Tool Reinforcement Learning](https://mlsanity.com/paper/2505.08617/) (2025)
- [Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models](https://mlsanity.com/paper/2408.15556/) (2024)
- [SimpleTIR: End-to-End Reinforcement Learning for Multi-Turn Tool-Integrated Reasoning](https://mlsanity.com/paper/2509.02479/) (2025)
- [DeepEyesV2: Toward Agentic Multimodal Model](https://mlsanity.com/paper/2511.05271/) (2025)
- [Mini-o3: Scaling Up Reasoning Patterns and Interaction Turns for Visual Search](https://mlsanity.com/paper/2509.07969/) (2025)
- [WebWatcher: Breaking New Frontier of Vision-Language Deep Research Agent](https://mlsanity.com/paper/2508.05748/) (2025)
- [Agentic Reasoning and Tool Integration for LLMs via Reinforcement Learning](https://mlsanity.com/paper/2505.01441/) (2025)

## Built on by (most cited first, 4 of 4)

- [LocalSearchBench: Benchmarking Agentic Search in Real-World Local Life Services](https://mlsanity.com/paper/2512.07436/) (2025)
- [VistaHop: Benchmarking Long-Horizon Visual DeepSearch](https://mlsanity.com/paper/2606.03273/) (2026)
- [TAPO: Tool-Aware Policy Optimization via Credit Transfer for Multimodal Search Agents](https://mlsanity.com/paper/2606.05784/) (2026)
- [Making Every Tool Call Count: Necessary Tool-Evidence Path Rewards for Agentic Vision-Language Models](https://mlsanity.com/paper/2609.03493/) (2026)

## Abstract (arXiv)

Recent VLM-based agents aim to replicate OpenAI O3's "thinking with images" via tool use, yet most open-source methods restrict inputs to a single image, limiting their applicability to real-world multi-image QA tasks. To address this gap, we propose IMAgent, an open-source visual agent trained with end-to-end reinforcement learning for fine-grained single/multi-image reasoning. During inference, VLMs tend to gradually neglect visual inputs; to mitigate this issue, we design two dedicated tools for visual reflection and verification, enabling the model to actively refocus attention on image content. Beyond that, we, for the first time, reveal how tool usage enhances agent performance from an attention perspective. Equipped with a carefully designed two-layer motion trajectory masking strategy and tool-use reward gain, IMAgent acquires an effective tool-use paradigm through pure reinforcement learning, eliminating the need for costly supervised fine-tuning data. To further unleash the inherent tool-usage potential of the base VLM and fill data gaps, we construct a challenging, visually enriched multi-image QA dataset via multi-agent system. Extensive experiments validate that IMAgent achieves SOTA performance across mainstream single and multi-image benchmarks, and our in-depth analysis offers actionable insights for the community. Code and data will be released soon.