MOEVE - Summarization: leaderboard
Metric: Mean overall summarization score x 100 (0-100) over four German datasets (Eur-Lex-Sum, Swiss Leading Decision Summarization, KIKC Summary, German Ministry Publications), combining BERTScore, SemScore, LLM-judged factual correctness and the German-output proportion; German prompts, default decoding, open-weight models on vLLM or Ollama at the quantization listed in the paper's model table, others by API; the paper prints only the top 10 of 39 evaluated models; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Mistral Large 3 | 78 |
| 2 | GPT-4o Mini | 77 |
| 3 | Phi-4 | 77 |
| 4 | Mistral Small 3.1 | 76.7 |
| 5 | Llama 3.3 70B | 76.6 |
| 6 | GPT-4o | 76.3 |
| 7 | DeepSeek R1 | 76.1 |
| 8 | GPT-OSS-120B | 75.8 |
Interactive version: theaggregate.ai/benchmark?slug=moeve-summarization · How It Works · Data refreshed daily, snapshot 2026-09-29.