EuroEval Slovene NLU - ScaLA SL: leaderboard

The "ScaLA SL" subtask of EuroEval Slovene NLU. Slovene-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. 204 models tracked.

Top models

#ModelScoreOverall rank
1Gemini 3 Pro (Preview)69.74#66
2Gemini 2.5 Flash (Thinking)64.71#245 (Gemini 2.5 Flash)
3Gemini 3.1 Flash Lite (Preview)62.19#194
4Gemini 2.5 Pro61.7#152
5Gemini 3 Flash (Preview) (Non-reasoning)60.93#82 (Gemini 3 Flash (Preview))
6GPT-559.76#98
7Claude Sonnet 4.659.72#90
8Gemini 3 Flash (Preview) (Thinking)58.98#82 (Gemini 3 Flash (Preview))
9Claude Sonnet 4.5 (Thinking)58.28#142 (Claude Sonnet 4.5)
10Gemini 2.5 Flash Lite (Thinking)58.26#422 (Gemini 2.5 Flash Lite)
11GPT-5.4 Mini (2026-03-17) (High)56.73#169 (GPT-5.4 Mini (2026-03-17))
12GPT-5.256.59#110
13GPT-5.4 Nano (2026-03-17) (High)53.89#266 (GPT-5.4 Nano (2026-03-17))
14GPT-5 Mini (2025-08-07)51.87#172
15Gemini 2.5 Flash (Non-reasoning)51.77#245 (Gemini 2.5 Flash)

No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.

Interactive version: theaggregate.ai/benchmark?slug=euroeval-slovene-nlu-scala-sl · How It Works · Data refreshed daily, snapshot 2026-10-20.