# Phi-4-reasoning-vision-15B Technical Report

arXiv: https://arxiv.org/abs/2603.03975 · HTML: https://mlsanity.com/paper/2603.03975/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 25 of its 40 references that are in the map.
- First posted on arXiv 4 Mar 2026.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)

## Builds on (most cited first, 25 of 25)

- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://mlsanity.com/paper/2301.12597/) (2023)
- [MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts](https://mlsanity.com/paper/2310.02255/) (2023)
- [ChartQA: A Benchmark for Question Answering about Charts with Visual and Logical Reasoning](https://mlsanity.com/paper/2203.10244/) (2022)
- [ShareGPT4V: Improving Large Multi-Modal Models with Better Captions](https://mlsanity.com/paper/2311.12793/) (2023)
- [Are We on the Right Way for Evaluating Large Vision-Language Models?](https://mlsanity.com/paper/2403.20330/) (2024)
- [A Diagram Is Worth A Dozen Images](https://mlsanity.com/paper/1603.07396/) (2016)
- [MathVerse: Does Your Multi-modal LLM Truly See the Diagrams in Visual Math Problems?](https://mlsanity.com/paper/2403.14624/) (2024)
- [LLaVA-NeXT-Interleave: Tackling Multi-image, Video, and 3D in Large Multimodal Models](https://mlsanity.com/paper/2407.07895/) (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)
- [SeeClick: Harnessing GUI Grounding for Advanced Visual GUI Agents](https://mlsanity.com/paper/2401.10935/) (2024)
- [A systematic study of the class imbalance problem in convolutional neural networks](https://mlsanity.com/paper/1710.05381/) (2017)
- [WildGuard: Open One-Stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMs](https://mlsanity.com/paper/2406.18495/) (2024)
- [The Hateful Memes Challenge: Detecting Hate Speech in Multimodal Memes](https://mlsanity.com/paper/2005.04790/) (2020)
- [V*: Guided Visual Search as a Core Mechanism in Multimodal LLMs](https://mlsanity.com/paper/2312.14135/) (2023)
- [OCRBench: On the Hidden Mystery of OCR in Large Multimodal Models](https://mlsanity.com/paper/2305.07895/) (2023)
- [Aguvis: Unified Pure Vision Agents for Autonomous GUI Interaction](https://mlsanity.com/paper/2412.04454/) (2024)
- [NVILA: Efficient Frontier Visual Language Models](https://mlsanity.com/paper/2412.04468/) (2024)
- [Simple data balancing achieves competitive worst-group-accuracy](https://mlsanity.com/paper/2110.14503/) (2021)
- [Safety Fine-Tuning at (Almost) No Cost: A Baseline for Vision Large Language Models](https://mlsanity.com/paper/2402.02207/) (2024)
- [Efficient Multimodal Learning from Data-centric Perspective](https://mlsanity.com/paper/2402.11530/) (2024)
- [Scaling Text-Rich Image Understanding via Code-Guided Synthetic Multimodal Data Generation](https://mlsanity.com/paper/2502.14846/) (2025)
- [VisOnlyQA: Large Vision Language Models Still Struggle with Visual Perception of Geometric Information](https://mlsanity.com/paper/2412.00947/) (2024)
- [VisualWebInstruct: Scaling up Multimodal Instruction Data through Web Search](https://mlsanity.com/paper/2503.10582/) (2025)
- [Why does Throwing Away Data Improve Worst-Group Error?](https://mlsanity.com/paper/2205.11672/) (2022)
- [Think in Safety: Unveiling and Mitigating Safety Alignment Collapse in Multimodal Large Reasoning Model](https://mlsanity.com/paper/2505.06538/) (2025)

## Built on by (most cited first, 10 of 10)

- [LatentRouter: Can We Choose the Right Multimodal Model Before Seeing Its Answer?](https://mlsanity.com/paper/2605.11301/) (2026)
- [Retrieval Heads Meet Vision: Uncovering How VLMs Locate and Extract Visual Information](https://mlsanity.com/paper/2608.27417/) (2026)
- [UltraVR: A Diagnostic Ultra-Resolution Image-VQA Benchmark for Evidence-Grounded Reasoning](https://mlsanity.com/paper/2606.05576/) (2026)
- [EinSort: Sorting is All We Need for Tensorizing LLM](https://mlsanity.com/paper/2606.08565/) (2026)
- [Knowing You at First Glance: Inferring Apparent Personality from Faces](https://mlsanity.com/paper/2607.14631/) (2026)
- [SciFigPlag-Bench: A Benchmark for Provenance-Aware Scientific Figure Plagiarism Detection](https://mlsanity.com/paper/2607.29124/) (2026)
- [When Safety Overrides Vision: Exploring Dynamics between Vision Influence and Safety Alignment in Vision-Language Models](https://mlsanity.com/paper/2608.18628/) (2026)
- [Rho: A Foundation for Efficiently Adaptable VLA Models](https://mlsanity.com/paper/2609.38164/) (2026)
- [Scenes as an Attack Surface: Uncovering Scene-Grounded Safety Failures in MLLMs](https://mlsanity.com/paper/2610.14094/) (2026)
- [ChemExplain: A Multimodal Benchmark for Explanatory Chemical Reasoning](https://mlsanity.com/paper/2610.17183/) (2026)

## Abstract (arXiv)

We present Phi-4-reasoning-vision-15B, a compact open-weight multimodal reasoning model, and share the motivations, design choices, experiments, and learnings that informed its development. Our goal is to contribute practical insight to the research community on building smaller, efficient multimodal reasoning models and to share the result of these learnings as an open-weight model that is good at common vision and language tasks and excels at scientific and mathematical reasoning and understanding user interfaces. Our contributions include demonstrating that careful architecture choices and rigorous data curation enable smaller, open-weight multimodal models to achieve competitive performance with significantly less training and inference-time compute and tokens. The most substantial improvements come from systematic filtering, error correction, and synthetic augmentation -- reinforcing that data quality remains the primary lever for model performance. Systematic ablations show that high-resolution, dynamic-resolution encoders yield consistent improvements, as accurate perception is a prerequisite for high-quality reasoning. Finally, a hybrid mix of reasoning and non-reasoning data with explicit mode tokens allows a single model to deliver fast direct answers for simpler tasks and chain-of-thought reasoning for complex problems.