FinRiskAtlas - Domain Knowledge - Risk & Compliance: leaderboard
Macro-averaged accuracy on the Risk & Compliance group of FinRiskAtlas domain-knowledge task families, a subset of the 42-family knowledge layer used for coarse screening of financial-review models (FinRiskAtlas, arXiv 2608.25325v1, Appendix Table 8). Percent; higher is better; zero-shot direct answers from 33 archived configurations.
Metric: Macro-Averaged Accuracy (%). Source: arxiv.org. Saturation forecast: Around February 2027. 33 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Kimi K3 | 78.11 | #35 |
| 2 | Gemini 3 Flash (Preview) | 77.24 | #90 |
| 3 | Qwen 3.7 Max | 77.15 | #85 |
| 4 | Claude Opus 4.7 | 76.55 | #51 |
| 5 | GPT-5.6 Sol | 76.3 | #22 |
| 6 | Kimi K2.5 | 75.23 | #152 |
| 7 | Ling-2.6-1T | 75.22 | #307 |
| 8 | GPT-5.6 Terra | 74.87 | #40 |
| 9 | GLM-5.2 | 74.65 | #80 |
| 10 | GPT-5.4 | 74.46 | #83 |
| 11 | GPT-5.4 Mini | 74.38 | #228 |
| 12 | Qwen3-Next-80B | 74.2 | #321 |
| 13 | Seed 2.0 Lite | 73.7 | #161 |
| 14 | GLM-5 | 73.53 | #149 |
| 15 | Qwen 3 235B A22B | 73.25 | #323 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=finriskatlas-domain-knowledge-risk-compliance · How It Works · Data refreshed daily, snapshot 2026-11-05.