LEDGER - Comprehensive KPI Extraction: leaderboard
Comprehensive KPI extraction on LEDGER (2026), a corpus of digitized corporate annual reports: the model extracts all 31 consolidated KPIs of each of the 494 test reports in one structured pass, and the score is recall over the 13,519 labelled KPI values with a 0.05 percent match tolerance. Higher is better.
Metric: Recall (%) of the 13,519 labelled KPI cells of the 494 test annual reports when the model extracts all 31 consolidated KPIs of a report in one structured pass (match tolerance 0.05%); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.6 27B | 77.2 | #341 |
| 2 | GPT-OSS-20B | 65.7 | #1147 |
| 3 | Nemotron 3 Nano | 65.5 | #1953 |
| 4 | Ministral 3 14B | 41.4 | #1845 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=ledger-comprehensive-kpi-extraction · How It Works · Data refreshed daily, snapshot 2026-11-13.