EuroEval German NLU - ScaLA DE: leaderboard
Metric: Linguistic acceptability Score (%). Source: euroeval.com. 325 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 74 |
| 2 | GPT-5.4 Mini (High) | 72.72 |
| 3 | Claude Sonnet 4.6 | 72.33 |
| 4 | Gemini 3.1 Flash Lite (Preview) | 71.73 |
| 5 | Gemma 4 31B (IT) | 68.08 |
| 6 | GPT-5.4 Mini (Medium) | 68.05 |
| 7 | GPT-5.2 | 66.66 |
| 8 | Llama 3.1 70B | 66.08 |
| 9 | gemma-3-27B-pt | 62.49 |
| 10 | GPT-5.4 Nano (Medium) | 62.04 |
| 11 | GPT-5.4 Nano (High) | 61.16 |
| 12 | Gemma 4 26B A4B (IT) | 61.08 |
| 13 | Llama 3.1 405B | 60.79 |
| 14 | Ministral 3 14B | 60.16 |
| 15 | Qwen 3.5 9B | 58.64 |
Interactive version: theaggregate.ai/benchmark?slug=euroeval-german-nlu-scala-de · How It Works · Data refreshed daily, snapshot 2026-09-05.