# R-4B: Incentivizing General-Purpose Auto-Thinking Capability in MLLMs via Bi-Mode Annealing and Reinforce Learning

arXiv: https://arxiv.org/abs/2508.21113 · HTML: https://mlsanity.com/paper/2508.21113/

## Key facts

- Cited by 18 papers in the map (data through 14 Oct 2026).
- Placed in Reinforcement learning for reasoning, one of 17 directions in LLM reasoning.
- Builds on 30 of its 39 references that are in the map.
- First posted on arXiv 28 Aug 2025.

## Placed in

- LLM reasoning / [Reinforcement learning for reasoning](https://mlsanity.com/lr/reinforcement-learning-for-reasoning/)
- Multimodal LLMs / vision-language models / [Embodied and 3D vision-language models](https://mlsanity.com/mllm/embodied-and-3d-vision-language-models/)

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

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [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)
- [DAPO: An Open-Source LLM Reinforcement Learning System at Scale](https://mlsanity.com/paper/2503.14476/) (2025)
- [LLaVA-OneVision: Easy Visual Task Transfer](https://mlsanity.com/paper/2408.03326/) (2024)
- [MMBench: Is Your Multi-modal Model an All-around Player?](https://mlsanity.com/paper/2307.06281/) (2023)
- [MMMU: A Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark for Expert AGI](https://mlsanity.com/paper/2311.16502/) (2023)
- [InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models](https://mlsanity.com/paper/2504.10479/) (2025)
- [Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling](https://mlsanity.com/paper/2412.05271/) (2024)
- [MM-Vet: Evaluating Large Multimodal Models for Integrated Capabilities](https://mlsanity.com/paper/2308.02490/) (2023)
- [SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features](https://mlsanity.com/paper/2502.14786/) (2025)
- [OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problems](https://mlsanity.com/paper/2402.14008/) (2024)
- [Are We on the Right Way for Evaluating Large Vision-Language Models?](https://mlsanity.com/paper/2403.20330/) (2024)
- [HallusionBench: An Advanced Diagnostic Suite for Entangled Language Hallucination and Visual Illusion in Large Vision-Language Models](https://mlsanity.com/paper/2310.14566/) (2023)
- [Measuring Multimodal Mathematical Reasoning with MATH-Vision Dataset](https://mlsanity.com/paper/2402.14804/) (2024)
- [MMMU-Pro: A More Robust Multi-discipline Multimodal Understanding Benchmark](https://mlsanity.com/paper/2409.02813/) (2024)
- [OCRBench: On the Hidden Mystery of OCR in Large Multimodal Models](https://mlsanity.com/paper/2305.07895/) (2023)
- [Seed1.5-VL Technical Report](https://mlsanity.com/paper/2505.07062/) (2025)
- [LogicVista: Multimodal LLM Logical Reasoning Benchmark in Visual Contexts](https://mlsanity.com/paper/2407.04973/) (2024)
- [AdaptThink: Reasoning Models Can Learn When to Think](https://mlsanity.com/paper/2505.13417/) (2025)
- [DynaMath: A Dynamic Visual Benchmark for Evaluating Mathematical Reasoning Robustness of Vision Language Models](https://mlsanity.com/paper/2411.00836/) (2024)
- [MiMo-VL Technical Report](https://mlsanity.com/paper/2506.03569/) (2025)
- [VisuLogic: A Benchmark for Evaluating Visual Reasoning in Multi-modal Large Language Models](https://mlsanity.com/paper/2504.15279/) (2025)
- [AdaCoT: Pareto-Optimal Adaptive Chain-of-Thought Triggering via Reinforcement Learning](https://mlsanity.com/paper/2505.11896/) (2025)
- [Ola: Pushing the Frontiers of Omni-Modal Language Model](https://mlsanity.com/paper/2502.04328/) (2025)
- [Learning When to Think: Shaping Adaptive Reasoning in R1-Style Models via Multi-Stage RL](https://mlsanity.com/paper/2505.10832/) (2025)
- [Kwai Keye-VL Technical Report](https://mlsanity.com/paper/2507.01949/) (2025)
- [RBench-V: A Primary Assessment for Visual Reasoning Models with Multi-modal Outputs](https://mlsanity.com/paper/2505.16770/) (2025)
- [KAT-V1: Kwai-AutoThink Technical Report](https://mlsanity.com/paper/2507.08297/) (2025)

## Built on by (most cited first, 18 of 18)

- [VideoAuto-R1: Video Auto Reasoning via Thinking Once, Answering Twice](https://mlsanity.com/paper/2601.05175/) (2026)
- [OralGPT-Omni: A Versatile Dental Multimodal Large Language Model](https://mlsanity.com/paper/2511.22055/) (2025)
- [CHURRO: Making History Readable with an Open-Weight Large Vision-Language Model for High-Accuracy, Low-Cost Historical Text Recognition](https://mlsanity.com/paper/2509.19768/) (2025)
- [Omni-AutoThink: Adaptive Multimodal Reasoning via Reinforcement Learning](https://mlsanity.com/paper/2512.03783/) (2025)
- [A Benchmark for Omni-Modal Reasoning in Long Videos](https://mlsanity.com/paper/2512.16978/) (2025)
- [Beyond Next-Token Alignment: Distilling Multimodal Large Language Models via Token Interactions](https://mlsanity.com/paper/2602.09483/) (2026)
- [Learning Adaptive Reasoning Paths for Efficient Visual Reasoning](https://mlsanity.com/paper/2604.14568/) (2026)
- [AndesVL Technical Report: An Efficient Mobile-side Multimodal Large Language Model](https://mlsanity.com/paper/2510.11496/) (2025)
- [OralAgent: Integrating Reasoning, Tools, and Knowledge for Interactive Dental Image Analysis](https://mlsanity.com/paper/2605.27378/) (2026)
- [Breaking Failure Cascades: Step-Aware Reinforcement Learning for Medical Multimodal Reasoning](https://mlsanity.com/paper/2606.31825/) (2026)
- [MM-THEBench: Do Reasoning MLLMs Think Reasonably?](https://mlsanity.com/paper/2601.22735/) (2026)
- [GroundCount: Grounding Vision-Language Models with Object Detection for Mitigating Counting Hallucinations](https://mlsanity.com/paper/2603.10978/) (2026)
- [DREAM-R: Multimodal Speculative Reasoning with RL-Based Refined Drafting, Precise Verification, and Fully Parallel Execution](https://mlsanity.com/paper/2605.28678/) (2026)
- [ADaPT: Token-Level Decoupling for Efficient Large Reasoning Models](https://mlsanity.com/paper/2606.19919/) (2026)
- [Switch-Reasoner: Learn When to Think in Multitask Mixtures via Reinforcement Learning](https://mlsanity.com/paper/2607.08572/) (2026)
- [AdaThinking-E: One-Token Entropy Regulation for Adaptive Thinking](https://mlsanity.com/paper/2608.26141/) (2026)
- [Counterfactual Tests for Measuring Chain-of-Thought Faithfulness in Visual Language Models](https://mlsanity.com/paper/2609.06704/) (2026)
- [Mixture-Trained Merging for Unified Multi-Objective Models](https://mlsanity.com/paper/2610.01238/) (2026)

## Abstract (arXiv)

Multimodal Large Language Models (MLLMs) equipped with step-by-step thinking capabilities have demonstrated remarkable performance on complex reasoning problems. However, this thinking process is redundant for simple problems solvable without complex reasoning. To address this inefficiency, we propose R-4B, an auto-thinking MLLM, which can adaptively decide when to think based on problem complexity. The central idea of R-4B is to empower the model with both thinking and non-thinking capabilities using bi-mode annealing, and apply Bi-mode Policy Optimization (BPO) to improve the model's accuracy in determining whether to activate the thinking process. Specifically, we first train the model on a carefully curated dataset spanning various topics, which contains samples from both thinking and non-thinking modes. Then it undergoes a second phase of training under an improved GRPO framework, where the policy model is forced to generate responses from both modes for each input query. Experimental results show that R-4B achieves state-of-the-art performance across 25 challenging benchmarks. It outperforms Qwen2.5-VL-7B in most tasks and achieves performance comparable to larger models such as Kimi-VL-A3B-Thinking-2506 (16B) on reasoning-intensive benchmarks with lower computational cost.