FinRiskAtlas - Evidence-Grounded Processing - Case Classification: leaderboard
FinRiskAtlas fixed-evidence review operation: routing a full case record to its procedure category from a fixed candidate set, scored by accuracy (FinRiskAtlas, arXiv 2608.25325v1, Appendix Table 9). Percent; higher is better. Reported for 33 archived model configurations under zero-shot direct-answer inference.
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 33 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Kimi K2.5 | 88.09 | #152 |
| 2 | GPT-5.6 Sol | 87.42 | #22 |
| 3 | Kimi K3 | 87.19 | #35 |
| 4 | GPT-5.4 | 86.29 | #83 |
| 5 | DeepSeek V4 Pro | 85.62 | #112 |
| 6 | GPT-5.4 Mini | 85.62 | #228 |
| 7 | GPT-5.6 Terra | 85.62 | #40 |
| 8 | Seed 2.0 Lite | 85.62 | #161 |
| 9 | Qwen 3.7 Max | 85.39 | #85 |
| 10 | Claude Opus 4.7 | 85.17 | #51 |
| 11 | Gemini 3 Flash (Preview) | 84.94 | #90 |
| 12 | GLM-5.2 | 84.94 | #80 |
| 13 | Qwen3-Next-80B | 84.94 | #321 |
| 14 | GLM-5 | 84.72 | #149 |
| 15 | GPT-5.6 Luna | 84.04 | #79 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=finriskatlas-evidence-grounded-processing-case-classification · How It Works · Data refreshed daily, snapshot 2026-11-05.