Flame Personal Finance - Evaluate: leaderboard
Textbook-grounded five-option multiple-choice problems at the Bloom Evaluate level (questions that ask the model to judge between alternatives or check a claim) from the 199-problem Flame Personal Finance benchmark (Islam et al., 2026). The paper's Bloom-level weighted accuracy in percent, zero-shot at temperature 0; higher is better.
Metric: Accuracy (%; the paper's Bloom-level weighted accuracy on the Evaluate problems of the 199-problem textbook-grounded Personal Finance multiple-choice set, zero-shot, temperature 0). Source: arxiv.org. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 100 | #91 |
| 2 | Claude Opus 4.6 | 96 | #100 |
| 3 | GPT-4.1 Mini | 88 | #775 |
| 4 | GPT-5.4 | 81 | #121 |
| 5 | Claude Haiku 4.5 | 77 | #568 |
| 6 | Gemini 2.5 Flash Lite | 75 | #958 |
| 7 | Qwen 3 32B | 58 | #960 |
| 8 | DeepSeek R1 Distill Qwen 32B | 52 | #1780 |
| 9 | Gemma 3 27B (IT) | 42 | #1306 |
| 10 | Qwen 3 8B | 40 | #1838 |
| 11 | DeepSeek R1 Distill Qwen 14B | 40 | #2594 |
| 12 | Gemma 3 12B (IT) | 35 | #1849 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=flame-personal-finance-evaluate · How It Works · Data refreshed daily, snapshot 2026-11-12.