HELM Finance - FinQA — leaderboard

Metric: Program Accuracy. Source: crfm.stanford.edu. 22 models tracked.

Top models

#ModelScore
1Gemini 1.5 Pro (002)58.7
2Llama 3.1 70B Instruct54.5
3Llama 3.2 90B Vision Instruct54.1
4Llama 3.1 405B Instruct54
5Gemini 1.5 Flash (001)53.5
6GPT-4o (2024-05-13)52.6
7GPT-4o Mini (2024-07-18)44.7
8Claude 3 Sonnet (20240229)42.8
9Qwen 2 72B Instruct42.4
10Llama 3 70B Instruct41.8
11Gemini 1.0 Pro (002)40.3
12GPT-3.5 Turbo (0125)37.7
13Claude 3 Opus (20240229)36.1
14Qwen1.5 Chat (72B)33.1
15Mistral 7B Instruct (v0.3)32.7

Interactive version: theaggregate.ai/benchmark?slug=helm-finance-finqa · How the rankings work · Data refreshed daily, snapshot 2026-07-22.