# Finch leaderboard

Agentic benchmark

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelchoice.dev/leaderboards/benchmarks/agentic-finch

Finch (FinWorkBench): Finance & Accounting across Spreadsheet-Centric Enterprise Workflows. Every model the catalog carries a published Finch value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | Agentic |
| Measure | Task score |
| Tasks | Spreadsheet-centric finance and accounting workflows |
| Difficulty | Enterprise finance work |
| Published by | [Finch: Benchmarking Finance & Accounting across Spreadsheet-Centric Enterprise Workflows](https://arxiv.org/abs/2512.13168) |

A benchmark of real finance and accounting workflows built around spreadsheets: creating, editing and analysing workbooks the way an analyst or accountant would.

## Finch ranking

5 models with a published Finch value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.

| Rank | Model | Provider | Task score | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [Kimi K3](https://modelchoice.dev/models/kimi-3) | Moonshot AI | 77.3 | Observed 2026-11-21 · source benchlm:models |
| 2 | [Mistral Large 4](https://modelchoice.dev/models/mistral-large-4) | Mistral | 73.3 | Observed 2026-11-21 · source benchlm:models |
| 3 | [DeepSeek V4 Pro](https://modelchoice.dev/models/deepseek-v4-pro) | DeepSeek | 67.4 | Observed 2026-11-21 · source benchlm:models |
| 3 | [DeepSeek V4 Pro 0813](https://modelchoice.dev/models/deepseek-v4-pro-max) | DeepSeek | 67.4 | Last good 2026-11-18 · source benchlm:models |
| 5 | [GLM-5.3](https://modelchoice.dev/models/glm-5-3) | Z.AI | 65.1 | Observed 2026-11-21 · source benchlm:models |

Rows are ordered by the value the source published, highest first. The source does not state whether a higher value is the better result, so this document does not either: for a benchmark that measures a rate of failure, read the table from the bottom.
