UniFinEval: Towards Unified Evaluation of Financial Multimodal Models across Text, Images and Videos

2026/01/09 by Zhi Yang, Lingfeng Zeng, Fangqi Lou +17 · 1 voice
Computer Science · Decision Sciences · Economics, Econometrics and Finance · #Benchmark (surveying) #Construct (python library) #Financial analysis #Financial modeling #Financial services #Financial statement #Multimodal Machine Learning Applications #Robustness (evolution) #Scope (computer science) #Stock Market Forecasting Methods #Topic Modeling #cs.AI #cs.CL #q-fin.GN

paper · pdf · doi:10.48550/arxiv.2601.22162

openalex publication_date 2026/01/09 · arxiv published 2026/01/09 · arxiv updated 2026/01/09 · openalex created_date 2026/02/03 · openalex updated_date 2026/07/28

Abstract

Multimodal large language models are playing an increasingly significant role in empowering the financial domain, however, the challenges they face, such as multimodal and high-density information and cross-modal multi-hop reasoning, go beyond the evaluation scope of existing multimodal benchmarks. To address this gap, we propose UniFinEval, the first unified multimodal benchmark designed for high-information-density financial environments, covering text, images, and videos. UniFinEval systematically constructs five core financial scenarios grounded in real-world financial systems: Financial Statement Auditing, Company Fundamental Reasoning, Industry Trend Insights, Financial Risk Sensing, and Asset Allocation Analysis. We manually construct a high-quality dataset consisting of 3,767 question-answer pairs in both chinese and english and systematically evaluate 10 mainstream MLLMs under Zero-Shot and CoT settings. Results show that Gemini-3-pro-preview achieves the best overall performance, yet still exhibits a substantial gap compared to financial experts. Further error analysis reveals systematic deficiencies in current models. UniFinEval aims to provide a systematic assessment of MLLMs' capabilities in fine-grained, high-information-density financial environments, thereby enhancing the robustness of MLLMs applications in real-world financial scenarios. Data and code are available at https://github.com/aifinlab/UniFinEval.

Citations

• FinMR: A Knowledge-Intensive Multimodal Benchmark for Advanced Financial Reasoning
• FinCap: Topic-Aligned Captions for Short-Form Financial YouTube Videos
• Unlocking Financial Insights: An advanced Multimodal Summarization with Multimodal Output Framework for Financial Advisory Videos
• Trading-R1: Financial Trading with LLM Reasoning via Reinforcement Learning
• MM-ARC: Multimodal Adaptive Routing of Capital with Robustness-Audited Strategy Pools
• A Survey on Open Dataset Search in the LLM Era: Retrospectives and Perspectives
• XFinBench: Benchmarking LLMs in Complex Financial Problem Solving and Reasoning
• VisFinEval: A Scenario-Driven Chinese Multimodal Benchmark for Holistic Financial Understanding
• FinMMR: Make Financial Numerical Reasoning More Multimodal, Comprehensive, and Challenging
• VideoConviction: A Multimodal Benchmark for Human Conviction and Stock Market Recommendations
• FinMME: Benchmark Dataset for Financial Multi-Modal Reasoning Evaluation
• DocTabQA: Answering Questions from Long Documents Using Tables
• MM-Vet v2: A Challenging Benchmark to Evaluate Large Multimodal Models for Integrated Capabilities
• Financial Statement Analysis with Large Language Models
• SEED-Bench-2-Plus: Benchmarking Multimodal Large Language Models with Text-Rich Visual Comprehension
• RiskLabs: Predicting Financial Risk Using Large Language Model based on Multimodal and Multi-Sources Data
• StockGPT: A GenAI Model for Stock Prediction and Trading
• AlphaFin: Benchmarking Financial Analysis with Retrieval-Augmented Stock-Chain Framework
• FinReport: Explainable Stock Earnings Forecasting via News Factor Analyzing Model
• A Multimodal Foundation Agent for Financial Trading: Tool-Augmented, Diversified, and Generalist
• DocFinQA: A Long-Context Financial Reasoning Dataset
• MMMU: A Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark for Expert AGI
• FinMem: A Performance-Enhanced LLM Trading Agent with Layered Memory and Character Design
• FinanceMath: Knowledge-Intensive Math Reasoning in Finance Domains
• Fin-Fact: A Benchmark Dataset for Multimodal Financial Fact Checking and Explanation Generation
• FinVis-GPT: A Multimodal Large Language Model for Financial Chart Analysis
• MultiFinBen: Benchmarking Large Language Models for Multilingual and Multimodal Financial Application
• FinChart-Bench: Benchmarking Financial Chart Comprehension in Vision-Language Models
• FAMMA: A Benchmark for Financial Domain Multilingual Multimodal Question Answering

Discussions

• Zhi Yang, et al.: UniFinEval: Towards Unified Evaluation of Financial Multimodal Models across Text, Images and Videos https://arxiv.org/abs/2601.22162 https://arxiv.org/pdf/2601.22162 https://arxiv.o [bsky, 0 points, 0 comments]

Related

• VisFinEval: A Scenario-Driven Chinese Multimodal Benchmark for Holistic Financial Understanding
• CFBenchmark-MM: Chinese Financial Assistant Benchmark for Multimodal Large Language Model
• MME-Finance: A Multimodal Finance Benchmark for Expert-level Understanding and Reasoning
• FinMME: Benchmark Dataset for Financial Multi-Modal Reasoning Evaluation
• MultiFinBen: Benchmarking Large Language Models for Multilingual and Multimodal Financial Application