Open FinLLM Reasoning - FinQA — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 26 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Fino1-14B | 74.18 |
| 2 | Qwen 2.5 72B Instruct | 73.38 |
| 3 | DeepSeek V3 | 73.2 |
| 4 | Qwen 2.5 32B Instruct | 73.11 |
| 5 | Fino1-8B | 73.03 |
| 6 | GPT-4o | 72.49 |
| 7 | Qwen2.5-Math-72B-Instruct | 69.74 |
| 8 | GPT-4.5 | 68.94 |
| 9 | Llama 3.3 70B Instruct | 68.15 |
| 10 | Qwen 2.5 14B Instruct | 67.44 |
| 11 | DeepSeek R1 Distill Llama 70B | 66.73 |
| 12 | DeepSeek R1 Distill Qwen 32B | 65.48 |
| 13 | DeepSeek R1 | 65.13 |
| 14 | DeepSeek R1 Distill Qwen 14B | 63.27 |
| 15 | Llama 3.1 70B Instruct | 63.18 |
Interactive version: theaggregate.ai/benchmark?slug=open-finllm-reasoning-finqa · How the rankings work · Data refreshed daily, snapshot 2026-07-22.