Pith. sign in

REVIEW 15 cited by

Reinforced MLLM: A Survey on RL-Based Reasoning in Multimodal Large Language Models

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2504.21277 v2 pith:F5GMLN42 submitted 2025-04-30 cs.AI

classification cs.AI
keywords reasoningmultimodallanguagelargemllmsmodelsrl-basedresearch
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

The application of reinforcement learning (RL) to enhance the reasoning capabilities of Multimodal Large Language Models (MLLMs) constitutes a rapidly advancing research area. While MLLMs extend Large Language Models (LLMs) to handle diverse modalities such as vision, audio, and video, enabling robust reasoning across multimodal inputs remains challenging. This paper provides a systematic review of recent advances in RL-based reasoning for MLLMs, covering key algorithmic designs, reward mechanism innovations, and practical applications. We highlight two main RL paradigms, value-model-free and value-model-based methods, and analyze how RL enhances reasoning abilities by optimizing reasoning trajectories and aligning multimodal information. Additionally, we provide an extensive overview of benchmark datasets, evaluation protocols, and current limitations, and propose future research directions to address challenges such as sparse rewards, inefficient cross-modal reasoning, and real-world deployment constraints. Our goal is to provide a comprehensive and structured guide to RL-based multimodal reasoning.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 15 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. SlowBA: An efficiency backdoor attack towards VLM-based GUI agents

    cs.CR 2026-03 conditional novelty 7.0 of 10

    SlowBA uses two-stage reward-level injection and pop-up triggers to make VLM GUI agents produce much longer, slower responses under attack while largely preserving task accuracy.

  2. Improving Generalization Robustness of Multimodal RLVR

    cs.AI 2026-08 conditional novelty 6.0 of 10

    PIRL, a trinary reward plus embedding-space adversarial consistency regularizer, reduces prompt-rewording degradation of RLVR-trained multimodal LLMs to about 1% average, versus about 3% for GRPO.

  3. Glance, Scrutinize, and Think: Advancing Video Anomaly Detection from Training-Free to Agentic Reasoning

    cs.AI 2026-08 conditional novelty 6.0 of 10

    A training-free glance-then-scrutinize pipeline and a reinforcement-learned tool-calling agent both improve joint temporal grounding and semantic understanding of video anomalies on a new benchmark.

  4. LongChart VQA: A Comprehensive Benchmark for MLLMs with Complex Multi-Chart Reasoning

    cs.CL 2026-08 reject novelty 6.0 of 10

    LongChart is a graph-consistent multi-chart VQA benchmark where 10 multimodal LLMs lose accuracy as question reasoning hops grow.

  5. Boosting RL-Based Visual Reasoning with Selective Adversarial Entropy Intervention

    cs.AI 2025-12 conditional novelty 6.0 of 10

    Selectively attacking the visual input with a gradient of response entropy during GRPO sampling increases policy entropy and improves visual reasoning accuracy by roughly 1–2 points.

  6. M2IO-R1: An Efficient RL-Enhanced Reasoning Framework for Multimodal Retrieval Augmented Multimodal Generation

    cs.IR 2025-08 conditional novelty 6.0 of 10

    M2IO-R1 uses GRPO reinforcement learning to train a 3B model that selects and places retrieved images into generated text answers, improving MRAMG quality on several benchmarks.

  7. VisualSphinx: Large-Scale Synthetic Vision Logic Puzzles for RL

    cs.CV 2025-05 conditional novelty 6.0 of 10

    LLM-derived rules, genetic expansion, and program-drawn images produce 660K visual logic puzzles; GRPO training on them lifts a VLM's accuracy on those puzzles and modestly on MathVista.

  8. Switch-Reasoner: Learn When to Think in Multitask Mixtures via Reinforcement Learning

    cs.CV 2026-07 conditional novelty 5.0 of 10

    A GRPO framework that treats thinking as a tool call and uses dual-level regulation so multimodal models learn when to reason versus answer directly.

  9. EMO-R3: Reflective Reinforcement Learning for Emotional Reasoning in Multimodal Large Language Models

    cs.AI 2026-02 conditional novelty 5.0 of 10

    EMO-R3, which combines a three-step emotional reasoning prompt with a reward for the model agreeing with its own image–emotion judgments, raises visual emotion-recognition accuracy by about one point over plain GRPO.

  10. Implicit Reasoning in Large Language Models: A Comprehensive Survey

    cs.CL 2025-09 conditional novelty 5.0 of 10

    A survey organizing implicit (silent) reasoning in LLMs into three execution paradigms, plus evidence, benchmarks, and challenges.

  11. WeThink: Toward General-purpose Vision-Language Reasoning via Reinforcement Learning

    cs.CV 2025-06 conditional novelty 5.0 of 10

    WeThink, a 120K-image QA dataset with AI-generated reasoning paths, combined with hybrid-reward reinforcement learning, improves a 7B vision-language model across 14 benchmarks.

  12. Argus Inspection: Do Multimodal Large Language Models Possess the Eye of Panoptes?

    cs.CV 2025-06 conditional novelty 5.0 of 10

    A new 1,430-item multimodal benchmark shows that leading multimodal LLMs rarely notice small visual traps needed for commonsense safety reasoning, with top scores near 0.46 on a scale whose maximum is about 0.97.

  13. Reinforcement Learning Meets Large Language Models: A Survey of Advancements and Applications Across the LLM Lifecycle

    cs.CL 2025-09 conditional novelty 3.0 of 10

    A survey that maps reinforcement learning methods, datasets, benchmarks, and open-source tools across the full training lifecycle of large language models, focusing on verifiable-reward reasoning.

  14. Reinforcement Fine-Tuning Powers Reasoning Capability of Multimodal Large Language Models

    cs.CL 2025-05 conditional novelty 2.0 of 10

    A survey-style position paper claims that reinforcement fine-tuning powers reasoning in multimodal LLMs, summarizing over a hundred recent works and proposing five future research directions.

  15. A Survey on Large Language Models for Mathematical Reasoning

    cs.AI 2025-06 conditional novelty 1.0 of 10

    Recent advances in LLM mathematical reasoning are organized into comprehension and generation phases, covering methods from prompting to test-time scaling.

Pith tools