EuroEval German NLU - ScaLA DE: leaderboard
The "ScaLA DE" subtask of EuroEval German NLU. German-language EuroEval natural-language-understanding suite, separating NLU task performance from the broader all-task EuroEval aggregate.
Metric: Matthews correlation coefficient (%). Source: euroeval.com. Saturation forecast: Around January 2027. 325 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 74 | #66 |
| 2 | GPT-5.4 Mini (2026-03-17) (High) | 72.72 | #169 (GPT-5.4 Mini (2026-03-17)) |
| 3 | Claude Sonnet 4.6 | 72.33 | #90 |
| 4 | Gemini 3.1 Flash Lite (Preview) | 71.73 | #194 |
| 5 | Gemini 3 Flash (Preview) (Thinking) | 70.89 | #82 (Gemini 3 Flash (Preview)) |
| 6 | Gemma 4 31B (IT) | 68.08 | #176 |
| 7 | GPT-5.2 | 66.66 | #110 |
| 8 | Gemini 3 Flash (Preview) (Non-reasoning) | 66.56 | #82 (Gemini 3 Flash (Preview)) |
| 9 | Llama 3.1 70B | 66.08 | #592 |
| 10 | EuroLLM-22B-Instruct-2512 | 65.65 | #701 |
| 11 | gemma-3-27B-pt | 62.49 | #502 |
| 12 | GPT-5.4 Nano (2026-03-17) (High) | 61.16 | #266 (GPT-5.4 Nano (2026-03-17)) |
| 13 | Gemma 4 26B A4B (IT) | 61.08 | #297 |
| 14 | Llama 3.1 405B | 60.79 | #466 |
| 15 | Qwen 3.5 9B | 58.64 | #379 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-german-nlu-scala-de · How It Works · Data refreshed daily, snapshot 2026-10-20.