EuroEval Slovene NLU - ScaLA SL: leaderboard
Metric: Linguistic acceptability Score (%). Source: euroeval.com. 204 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 69.74 |
| 2 | Gemini 2.5 Flash (Thinking) | 64.71 |
| 3 | Gemini 3.1 Flash Lite (Preview) | 62.19 |
| 4 | Gemini 2.5 Pro | 61.7 |
| 5 | GPT-5 | 59.76 |
| 6 | Claude Sonnet 4.6 | 59.72 |
| 7 | GPT-5.4 Mini (Medium) | 58.72 |
| 8 | Claude Sonnet 4.5 (Thinking) | 58.28 |
| 9 | GPT-5.4 Mini (High) | 56.73 |
| 10 | GPT-5.2 | 56.59 |
| 11 | GPT-5.4 Nano (High) | 53.89 |
| 12 | GPT-5 Mini | 51.87 |
| 13 | Gemma 4 31B (IT) | 51.75 |
| 14 | GPT-5.4 Nano (Medium) | 51.14 |
| 15 | Gemma 3 27B (IT) | 50.53 |
Interactive version: theaggregate.ai/benchmark?slug=euroeval-slovene-nlu-scala-sl · How It Works · Data refreshed daily, snapshot 2026-09-05.