# Finch: Benchmarking Finance & Accounting across Spreadsheet-Centric Enterprise Workflows

arXiv: https://arxiv.org/abs/2512.13168 · HTML: https://mlsanity.com/paper/2512.13168/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- Builds on 37 of its 79 references that are in the map.
- First posted on arXiv 15 Dec 2025.

## Builds on (most cited first, 30 of 37)

- [DocVQA: A Dataset for VQA on Document Images](https://mlsanity.com/paper/2007.00398/) (2020)
- [FinQA: A Dataset of Numerical Reasoning over Financial Data](https://mlsanity.com/paper/2109.00122/) (2021)
- [FinanceBench: A New Benchmark for Financial Question Answering](https://mlsanity.com/paper/2311.11944/) (2023)
- [CharXiv: Charting Gaps in Realistic Chart Understanding in Multimodal LLMs](https://mlsanity.com/paper/2406.18521/) (2024)
- [GDPval: Evaluating AI Model Performance on Real-World Economically Valuable Tasks](https://mlsanity.com/paper/2510.04374/) (2025)
- [FinBen: A Holistic Financial Benchmark for Large Language Models](https://mlsanity.com/paper/2402.12659/) (2024)
- [SpreadsheetBench: Towards Challenging Real World Spreadsheet Manipulation](https://mlsanity.com/paper/2406.14991/) (2024)
- [TableBank: A Benchmark Dataset for Table Detection and Recognition](https://mlsanity.com/paper/1903.01949/) (2019)
- [TUTA: Tree-based Transformers for Generally Structured Table Pre-training](https://mlsanity.com/paper/2010.12537/) (2020)
- [OfficeBench: Benchmarking Language Agents across Multiple Applications for Office Automation](https://mlsanity.com/paper/2407.19056/) (2024)
- [WideSearch: Benchmarking Agentic Broad Info-Seeking](https://mlsanity.com/paper/2508.07999/) (2025)
- [Finance Agent Benchmark: Benchmarking LLMs on Real-world Financial Research Tasks](https://mlsanity.com/paper/2508.00828/) (2025)
- [INVESTORBENCH: A Benchmark for Financial Decision-Making Tasks with LLM-based Agent](https://mlsanity.com/paper/2412.18174/) (2024)
- [SheetCopilot: Bringing Software Productivity to the Next Level through Large Language Models](https://mlsanity.com/paper/2305.19308/) (2023)
- [OdysseyBench: Evaluating LLM Agents on Long-Horizon Complex Office Application Workflows](https://mlsanity.com/paper/2508.09124/) (2025)
- [DABstep: Data Agent Benchmark for Multi-step Reasoning](https://mlsanity.com/paper/2506.23719/) (2025)
- [Multimodal Table Understanding](https://mlsanity.com/paper/2406.08100/) (2024)
- [FinAgentBench: A Benchmark Dataset for Agentic Retrieval in Financial Question Answering](https://mlsanity.com/paper/2508.14052/) (2025)
- [FinSearchComp: Towards a Realistic, Expert-Level Evaluation of Financial Search and Reasoning](https://mlsanity.com/paper/2509.13160/) (2025)
- [FORTAP: Using Formulas for Numerical-Reasoning-Aware Table Pretraining](https://mlsanity.com/paper/2109.07323/) (2021)
- [SheetAgent: Towards A Generalist Agent for Spreadsheet Reasoning and Manipulation via Large Language Models](https://mlsanity.com/paper/2403.03636/) (2024)
- [Tapilot-Crossing: Benchmarking and Evolving LLMs Towards Interactive Data Analysis Agents](https://mlsanity.com/paper/2403.05307/) (2024)
- [SpreadsheetLLM: Encoding Spreadsheets for Large Language Models](https://mlsanity.com/paper/2407.09025/) (2024)
- [Auto-Tables: Synthesizing Multi-Step Transformations to Relationalize Tables without Using Examples](https://mlsanity.com/paper/2307.14565/) (2023)
- [FinAuditing: A Financial Taxonomy-Structured Multi-Document Benchmark for Evaluating LLMs](https://mlsanity.com/paper/2510.08886/) (2025)
- [MiMoTable: A Multi-scale Spreadsheet Benchmark with Meta Operations for Table Reasoning](https://mlsanity.com/paper/2412.11711/) (2024)
- [TableSense: Spreadsheet Table Detection with Convolutional Neural Networks](https://mlsanity.com/paper/2106.13500/) (2021)
- [Auto-Formula: Recommend Formulas in Spreadsheets using Contrastive Learning for Table Representations](https://mlsanity.com/paper/2404.12608/) (2024)
- [RealHiTBench: A Comprehensive Realistic Hierarchical Table Benchmark for Evaluating LLM-Based Table Analysis](https://mlsanity.com/paper/2506.13405/) (2025)
- [XFinBench: Benchmarking LLMs in Complex Financial Problem Solving and Reasoning](https://mlsanity.com/paper/2508.15861/) (2025)

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

- [FutureX-Pro: Extending Future Prediction to High-Value Vertical Domains](https://mlsanity.com/paper/2601.12259/) (2026)
- [Beyond Rows to Reasoning: Agentic Retrieval for Multimodal Spreadsheet Understanding and Editing](https://mlsanity.com/paper/2603.06503/) (2026)
- [Herculean: An Agentic Benchmark for Financial Intelligence](https://mlsanity.com/paper/2605.14355/) (2026)
- [DocOps: A Verifiable Benchmark for Autonomous Agents in Complex Document Operations](https://mlsanity.com/paper/2607.19865/) (2026)
- [IBIB: A Protocol for Measuring Enterprise AI Systems by Serving Route, Not Model Identifier](https://mlsanity.com/paper/2609.10494/) (2026)
- [FinRCA-Bench: Benchmarking Evidence Retrieval and Reasoning for Financial AI Systems](https://mlsanity.com/paper/2608.18534/) (2026)
- [FinExam-10K: When Retrieval Helps Financial Reasoning?](https://mlsanity.com/paper/2608.28155/) (2026)
- [FinCUABuild: Can Agents Build Reliable Benchmarks for Dynamic Financial Computer Use?](https://mlsanity.com/paper/2609.07603/) (2026)
- [Debiasing as a Measurement Intervention: Calibrated Ties and Resolution Loss in LLM-as-a-Judge Evaluation](https://mlsanity.com/paper/2609.12439/) (2026)
- [Knowledge or Calculator? Decomposing the Skill Premium in Verifiable Financial Agent Workflows](https://mlsanity.com/paper/2610.03564/) (2026)

## Abstract (arXiv)

We introduce FinWorkBench (a.k.a. Finch) for evaluating AI agents on real-world, enterprise-grade finance and accounting workflows that interleave data entry, structuring, formatting, web search, cross-file retrieval, calculation, modeling, validation, translation, visualization, and reporting. Finch is sourced from authentic enterprise workspaces from Enron (15,000 files and 500,000 emails) and other financial institutions, covering the period 2000--2025 and preserving the in-the-wild messiness of multimodal artifacts such as tables and charts across diverse domains including budgeting, trading, asset management, and operational management. We propose a workflow construction process that combines LLM-assisted mining of workflows from authentic enterprise environments with expert annotation: (1) LLM-assisted, expert-verified derivation of workflows from real-world email threads and spreadsheet version histories, and (2) meticulous annotation requiring over 700 hours of expert effort. This yields 172 composite workflows with 384 tasks, involving 1,710 spreadsheets with 27 million cells, along with PDFs and other artifacts, capturing the intrinsically messy, long-horizon, knowledge-intensive, and collaborative nature of real-world enterprise work. We conduct both human and automated evaluations of frontier AI systems, including GPT-5.1, Claude Sonnet 4.5, Claude Opus 4.5, Gemini 3 Pro, Grok 4, and Qwen 3 Max. Under human evaluation, GPT-5.1 Pro spends an average of 16.8 minutes per workflow yet passes only 38.4% of workflows. Comprehensive case studies further surface the challenges that real-world enterprise workflows pose for AI agents.