EuroEval Icelandic NLU - ScaLA IS: leaderboard

The "ScaLA IS" subtask of EuroEval Icelandic NLU. Icelandic-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 August 2028. 398 models tracked.

Top models

#ModelScoreOverall rank
1Gemini 3 Pro (Preview)69.88#66
2Claude 3.7 Sonnet (20250219) (Thinking)62.13#201 (Claude 3.7 Sonnet (20250219))
3Gemini 3 Flash (Preview) (Non-reasoning)61.88#82 (Gemini 3 Flash (Preview))
4GPT-5 (2025-08-07) (High)59.65
5Gemini 3.1 Flash Lite (Preview)59.17#194
6GPT-557.81#98
7Claude Sonnet 4.5 (Thinking)56.61#142 (Claude Sonnet 4.5)
8Gemini 3 Flash (Preview) (Thinking)56.16#82 (Gemini 3 Flash (Preview))
9O3 (2025-04-16)55.32#121
10Gemini 2.5 Flash (Thinking)53.96#245 (Gemini 2.5 Flash)
11O4 Mini (2025-04-16)53.35#179
12Gemini 2.5 Pro53.03#152
13Gemini 2.5 Pro (Preview 03-25)51.46#123
14GPT-5 Mini (2025-08-07) (High)50.58#172 (GPT-5 Mini (2025-08-07))
15GPT-4.150.04#248

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

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