BlogLLM Benchmark Scores as Published, Side by Side

FinWorkBench: Results as Published

Every result the catalog holds for FinWorkBench: 7 models from 5 developers, read from 1 page. Each number is as printed by the company named beside it; nothing is ranked or rescored. Models are listed by developer, newest first.

Area Knowledge WorkMaintainer Haoyu Dong and co-authorsIntroduced 2025Data version 2026103001, as of 2026-10-30

FinWorkBench, also called Finch, tests agents on finance and accounting work taken from real company files. A workflow can mix data entry, structuring, formatting, web search, cross-file retrieval, calculation, modeling, validation, translation, visualization, and reporting. The release has 172 composite workflows and 384 tasks, including 1,710 spreadsheets with 27 million cells, plus PDFs and other files, drawn from Enron archives and other financial institutions. Human judges mark whether a workflow passes, and the paper also runs an automated evaluation.

FinWorkBench (Finch)

ModelValuePrinted bySetup as printed
GPT-6 AstraOpenAI, released 2026-09-0357%Mistral AI, Mistral Large 4 (preview)checked 2026-10-15-
DeepSeek-V4-Pro-0813DeepSeek67.4%Mistral AI, Mistral Large 4 (preview)checked 2026-10-15-
Kimi K3Moonshot AI, released 2026-07-1677.3%Mistral AI, Mistral Large 4 (preview)checked 2026-10-15-
GLM-5.3Z.ai, released 2026-08-1865.1%Mistral AI, Mistral Large 4 (preview)checked 2026-10-15-
GLM-5.2Z.ai, released 2026-06-1661%Mistral AI, Mistral Large 4 (preview)checked 2026-10-15-
Mistral Large 4 (preview)Mistral AI, released 2026-10-0667.4%Mistral AI, Mistral Large 4 (preview)checked 2026-10-15-
Mistral Medium 3.5Mistral AI, released 2026-05-2236.7%Mistral AI, Mistral Large 4 (preview)checked 2026-10-15-

Every number here is read from the same catalog file as the benchmark table, data version 2026103001, and is the number the table shows for that model and test. Where one page printed a model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.