ConvFinQA
Financial calculations across a sequence of related questions.
Fin-R1 · 2025 · published research
Evaluation setup
Author-reported Fin-R1 evaluation results. These scores use a different study setup from the Fino1 views. They must not be combined into a single overall model rank.
Reading this comparison
Higher values indicate better performance under the named study's scoring rules. Published model names are retained. These are historical snapshots, not tests of later versions with similar names.
Boundaries
A benchmark answer score does not establish a model's readiness to deliver a financial diligence engagement. It does not measure CPA sign-off, evidence retention, access controls, or performance on your transaction documents.
Provenance
Transcribed from the linked primary source on September 12, 2026. Evaluation was performed by the source authors, not Finsider. The downloadable CSV includes each value, configuration, and source URL.
Performance Comparison
Reported score (%)
Fin-R1
85.00
DeepSeek-R1
82.00
Fin-R1-SFT
81.00
Qwen-2.5-32B-Instruct
78.00
Qwen-2.5-14B-Instruct
77.00
DeepSeek-R1-Distill-Llama-70B
74.00
DeepSeek-R1-Distill-Qwen-14B
73.00
DeepSeek-R1-Distill-Qwen-32B
72.00
Qwen-2.5-7B-Instruct
66.00
DeepSeek-R1-Distill-Qwen-7B
62.00
All ten models in the source comparison.
Rank uses displayed scores, and ties share rank. Bars use a 0–100 scale. The source table provides no confidence intervals, so none are drawn.
Source: Fin-R1 · 2025