FrontierFinance - Financial Data & Modeling: leaderboard
The 70 FrontierFinance queries on financial data retrieval and modeling, answered in the Finance Agent v2 harness. Graded against expert rubric items by a three-judge majority vote; the score is the mean per-query share of rubric items satisfied (%), and higher is better.
Metric: Rubric qualification rate (%, macro average over queries, three-judge majority vote). Source: research.samaya.ai. Saturation forecast: Around May 2028. 16 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5.1 (High) | 60.94 | #14 (Claude Fable 5.1) |
| 2 | Claude Opus 5.5 (High) | 57.25 | #5 (Claude Opus 5.5) |
| 3 | Gemini 3.6 Flash (Medium) | 56.23 | #72 (Gemini 3.6 Flash) |
| 4 | Gemini 3.8 Flash (Medium) | 56.16 | #36 (Gemini 3.8 Flash) |
| 5 | Muse Spark Open (xHigh) | 55.83 | #79 (Muse Spark Open) |
| 6 | Claude Fable 5 (High) | 55.56 | #26 (Claude Fable 5) |
| 7 | Muse Wonder | 54.57 | #27 |
| 8 | Claude Opus 4.8 (High) | 50.96 | #56 (Claude Opus 4.8) |
| 9 | GPT-5.6 Sol (Medium) | 48.95 | #29 (GPT-5.6 Sol) |
| 10 | GPT-5.5 (Medium) | 47.6 | #48 (GPT-5.5) |
| 11 | GLM-5.2 (Max) | 47.53 | #137 (GLM-5.2) |
| 12 | Ling-3.0-flash-fin | 46.81 | #289 |
| 13 | Kimi K3 (Max) | 46.51 | #53 (Kimi K3) |
| 14 | DeepSeek V4 Pro (High) | 45.3 | #181 (DeepSeek V4 Pro) |
| 15 | Gemini 3.1 Pro (Preview) (High) | 31.36 | #91 (Gemini 3.1 Pro (Preview)) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #6 GPT-6 Astra.
Interactive version: theaggregate.ai/benchmark?slug=frontierfinance-financial-data-modeling · How It Works · Data refreshed daily, snapshot 2026-11-12.