German Public Sector Benchmark: leaderboard
Party-position classification accuracy from the paper's top-ten model table for German public-sector applications. Scores are fractions from 0 to 1, retaining the published three-decimal precision.
Metric: Party-Position Classification Accuracy (0-1). Source: arxiv.org. Saturation forecast: Around April 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o Mini | 0.67 | #599 |
| 2 | DeepSeek R1 | 0.67 | #251 |
| 3 | GPT-OSS-120B | 0.67 | #341 |
| 4 | Llama 3.3 70B | 0.66 | #579 |
| 5 | Mistral Small 3.1 | 0.65 | #610 |
| 6 | GPT-4o | 0.62 | #343 |
| 7 | GPT-OSS-20B | 0.62 | #518 |
| 8 | Mistral Large 3 | 0.61 | #401 |
| 9 | DeepSeek R1 Distill Qwen 32B | 0.6 | #655 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=german-public-sector-benchmark · How It Works · Data refreshed daily, snapshot 2026-10-20.