HELM Finance - FinQA — leaderboard
Metric: Program Accuracy. Source: crfm.stanford.edu. 22 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 1.5 Pro (002) | 58.7 |
| 2 | Llama 3.1 70B Instruct | 54.5 |
| 3 | Llama 3.2 90B Vision Instruct | 54.1 |
| 4 | Llama 3.1 405B Instruct | 54 |
| 5 | Gemini 1.5 Flash (001) | 53.5 |
| 6 | GPT-4o (2024-05-13) | 52.6 |
| 7 | GPT-4o Mini (2024-07-18) | 44.7 |
| 8 | Claude 3 Sonnet (20240229) | 42.8 |
| 9 | Qwen 2 72B Instruct | 42.4 |
| 10 | Llama 3 70B Instruct | 41.8 |
| 11 | Gemini 1.0 Pro (002) | 40.3 |
| 12 | GPT-3.5 Turbo (0125) | 37.7 |
| 13 | Claude 3 Opus (20240229) | 36.1 |
| 14 | Qwen1.5 Chat (72B) | 33.1 |
| 15 | Mistral 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.