FinanceQA - Conceptual
Finance & Accounting · 2025-01-29
The conceptual FinanceQA component: 64 questions on financial-analysis, accounting and valuation concepts drawn from textbooks and annotators' firms, asked without document context. Answers are graded as exact correct matches.
Top models (higher is better)
| Model | Score |
|---|---|
| o3 | 76.6 |
| Grok 4 | 75.0 |
| o4-mini | 75.0 |
| Llama 4 Maverick Instruct | 73.4 |
| Llama 3.1 Nemotron Ultra 253B V1 | 71.9 |
| Phi-4-reasoning-plus | 70.3 |
| Grok 3 (API) | 68.8 |
| Gemini 2.5 Pro | 67.2 |
| Opus 4 | 67.2 |
| Sonnet 4 | 65.6 |
| GPT-4.1 Mini | 65.6 |
| DeepSeek-R1-0528 | 65.1 |
| QwQ-32B | 64.1 |
| Qwen3-30B-A3B | 59.4 |
| Magistral Medium 1.0 | 53.1 |
| Gemini 2.5 Flash | 51.6 |
| Kimi K2 Instruct | 50.0 |
| Command A | 35.9 |
| Nova Pro | 26.6 |