HELM Finance - FinanceBench: leaderboard
HELM Finance run of FinanceBench, open-book financial question answering scored by correct answers.
Metric: Correct Answer. Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 22 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude 3 Opus (20240229) | 70 | #412 |
| 2 | GPT-4o (2024-05-13) | 66.43 | #325 |
| 3 | Claude 3.5 Sonnet (20240620) | 65 | #299 |
| 4 | Claude 3 Sonnet (20240229) | 65 | #677 |
| 5 | Llama 3.1 405B Instruct | 62.86 | #457 |
| 6 | Llama 3 70B Instruct | 60.71 | #634 |
| 7 | GPT-3.5 Turbo (0125) | 60 | #984 |
| 8 | Gemini 1.5 Pro (002) | 57.86 | #349 |
| 9 | Gemini 1.5 Flash (001) | 57.14 | #577 |
| 10 | Llama 3.2 90B Vision Instruct | 55.71 | #622 |
| 11 | Qwen 2 72B Instruct | 55 | #576 |
| 12 | GPT-4o Mini (2024-07-18) | 54.29 | #558 |
| 13 | Llama 3.1 70B Instruct | 53.57 | #561 |
| 14 | Qwen 1.5 72B Chat | 52.86 | #776 |
| 15 | Mixtral 8x22B Instruct | 52.14 | #728 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-finance-financebench · How It Works · Data refreshed daily, snapshot 2026-10-20.