EuroEval German NLU - Sb10K: leaderboard
The "Sb10K" 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: Estimated already saturated. 326 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 Mini (2025-08-07) | 64.7 | #172 |
| 2 | GPT-5 Mini (2025-08-07) (High) | 64.63 | #172 (GPT-5 Mini (2025-08-07)) |
| 3 | O3 (2025-04-16) | 64.04 | #121 |
| 4 | Mistral Small 3.2 | 62.28 | #605 |
| 5 | Mistral Small 3 | 62.27 | #627 |
| 6 | Gemini 2.5 Flash (Preview 04-17) | 61.7 | #237 |
| 7 | Qwen 3 235B A22B 2507 FP8 (Thinking) | 61.11 | #246 (Qwen 3 235B A22B 2507 FP8) |
| 8 | Mistral Small 3.1 | 60.32 | #610 |
| 9 | cogito-v1-preview-qwen-14B | 60.3 | |
| 10 | Llama 3.1 405B Instruct FP8 | 60.23 | #389 |
| 11 | O4 Mini (2025-04-16) | 60.22 | #179 |
| 12 | Grok 3 Mini | 60.13 | #281 |
| 13 | QwQ-32B | 59.76 | #421 |
| 14 | Reka Flash 3 | 59.6 | #619 |
| 15 | NeuralBeagle14-7B | 59.6 | #850 |
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-sb10k · How It Works · Data refreshed daily, snapshot 2026-10-20.