EuroEval Ukrainian NLU - Cross Domain UK Reviews: leaderboard

The "Cross Domain UK Reviews" subtask of EuroEval Ukrainian NLU. Ukrainian-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. 201 models tracked.

Top models

#ModelScoreOverall rank
1gemma-3-12B-pt65.19#686
2lapa-v0.1.2-instruct65.1#672
3Llama-3.1-Tulu-3-8B-SFT64.17#985
4MamayLM-Gemma-3-12B-IT-v1.064.08#584
5Gemma 3 12B (IT)63.71#664
6Llama 3.3 70B Instruct63.5#534
7Qwen 2.5 72B Instruct63.38#454
8Gemini 3 Pro (Preview)63.25#66
9gemma-3-4B-pt Base62.56#1134
10Gemma 3n E4B (IT)62.56#749
11gemma-3-27B-pt62.28#502
12Qwen 3 8B (Non-reasoning)62.26#674 (Qwen 3 8B)
13Gemma 3 27B (IT)62.1#523
14Qwen 3 14B (Non-reasoning)61.67#537 (Qwen 3 14B)
15Olmo-3.1-32B-Instruct-DPO61.48#845

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

Interactive version: theaggregate.ai/benchmark?slug=euroeval-ukrainian-nlu-cross-domain-uk-reviews · How It Works · Data refreshed daily, snapshot 2026-10-20.