Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Survey of Large Language Models in Finance (FinLLMs)

Jean Lee, Nicholas Stevens, Soyeon Caren Han, Minseok Song

TL;DR

This survey analyzes the emergence of FinLLMs by tracing the evolution from general-domain LMs to finance-specific models, outlining training techniques (continual, domain-specific, mixed-domain, and instruction-tuned) and evaluating them on six core financial benchmarks. It highlights that mixed-domain FinPLMs often excel at sentiment, classification, and NER tasks, while task-specific models/QA and SMP performance remains stronger for specialized approaches. The paper introduces eight advanced financial NLP tasks with corresponding datasets and discusses practical challenges such as privacy, efficiency, and evaluation alignment, proposing retrieval-augmented methods and richer datasets to push the field forward. Overall, it provides a roadmap for developing robust, trustworthy FinLLMs with broader real-world impact in finance, including robo-advisory, trading, and document understanding.

Abstract

Large Language Models (LLMs) have shown remarkable capabilities across a wide variety of Natural Language Processing (NLP) tasks and have attracted attention from multiple domains, including financial services. Despite the extensive research into general-domain LLMs, and their immense potential in finance, Financial LLM (FinLLM) research remains limited. This survey provides a comprehensive overview of FinLLMs, including their history, techniques, performance, and opportunities and challenges. Firstly, we present a chronological overview of general-domain Pre-trained Language Models (PLMs) through to current FinLLMs, including the GPT-series, selected open-source LLMs, and financial LMs. Secondly, we compare five techniques used across financial PLMs and FinLLMs, including training methods, training data, and fine-tuning methods. Thirdly, we summarize the performance evaluations of six benchmark tasks and datasets. In addition, we provide eight advanced financial NLP tasks and datasets for developing more sophisticated FinLLMs. Finally, we discuss the opportunities and the challenges facing FinLLMs, such as hallucination, privacy, and efficiency. To support AI research in finance, we compile a collection of accessible datasets and evaluation benchmarks on GitHub.

A Survey of Large Language Models in Finance (FinLLMs)

TL;DR

This survey analyzes the emergence of FinLLMs by tracing the evolution from general-domain LMs to finance-specific models, outlining training techniques (continual, domain-specific, mixed-domain, and instruction-tuned) and evaluating them on six core financial benchmarks. It highlights that mixed-domain FinPLMs often excel at sentiment, classification, and NER tasks, while task-specific models/QA and SMP performance remains stronger for specialized approaches. The paper introduces eight advanced financial NLP tasks with corresponding datasets and discusses practical challenges such as privacy, efficiency, and evaluation alignment, proposing retrieval-augmented methods and richer datasets to push the field forward. Overall, it provides a roadmap for developing robust, trustworthy FinLLMs with broader real-world impact in finance, including robo-advisory, trading, and document understanding.

Abstract

Large Language Models (LLMs) have shown remarkable capabilities across a wide variety of Natural Language Processing (NLP) tasks and have attracted attention from multiple domains, including financial services. Despite the extensive research into general-domain LLMs, and their immense potential in finance, Financial LLM (FinLLM) research remains limited. This survey provides a comprehensive overview of FinLLMs, including their history, techniques, performance, and opportunities and challenges. Firstly, we present a chronological overview of general-domain Pre-trained Language Models (PLMs) through to current FinLLMs, including the GPT-series, selected open-source LLMs, and financial LMs. Secondly, we compare five techniques used across financial PLMs and FinLLMs, including training methods, training data, and fine-tuning methods. Thirdly, we summarize the performance evaluations of six benchmark tasks and datasets. In addition, we provide eight advanced financial NLP tasks and datasets for developing more sophisticated FinLLMs. Finally, we discuss the opportunities and the challenges facing FinLLMs, such as hallucination, privacy, and efficiency. To support AI research in finance, we compile a collection of accessible datasets and evaluation benchmarks on GitHub.
Paper Structure (23 sections, 2 figures, 1 table)

This paper contains 23 sections, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. Evolution Trends: from General to Finance
  3. General-domain LMs
  4. GPT-Series
  5. Open-source LLMs
  6. Financial-domain LMs
  7. Techniques: from FinPLMs to FinLLMs
  8. Continual Pre-training
  9. Domain-Specific Pre-training from Scratch
  10. Mixed-Domain Pre-training
  11. Mixed-Domain LLM with Prompt Engineering
  12. Instruction Fine-tuned LLM with Prompt Engineering
  13. Evaluation: Benchmark Tasks and Datasets
  14. Sentiment Analysis (SA)
  15. Text Classification (TC)
  16. ...and 8 more sections

Figures (2)

  • Figure 1: Timeline showing the evolution of selected PLM/LLM releases from the general domain to the financial domain.
  • Figure 2: Comparison of techniques used in financial LMs: from FinPLMs to FinLLMs.