QFBench: leaderboard
Quantitative-finance agent benchmark with verifiable code-execution tasks for derivatives, risk, factors, and trading calculations.
Metric: Pass@1 (%). Source: qfbench.com. Source: benchmarklist.com. Saturation forecast: Around January 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 61.7 | #31 |
| 2 | Claude Opus 4.7 | 61.2 | #49 |
| 3 | GPT-5.3 Codex | 60.8 | #69 |
| 4 | Claude Opus 4.6 | 59.2 | #63 |
| 5 | GPT-5.4 | 57.5 | #79 |
| 6 | GPT-5.4 Mini | 57.1 | #212 |
| 7 | Claude Sonnet 4.6 | 56.3 | #90 |
| 8 | Claude Sonnet 4.5 | 46.2 | #142 |
| 9 | Claude Haiku 4.5 | 20.8 | #280 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=qfbench · How It Works · Data refreshed daily, snapshot 2026-10-20.