EuroEval Croatian NLU - MMS HR: leaderboard
The "MMS HR" subtask of EuroEval Croatian NLU. Croatian-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 2028. 212 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.1 70B | 48.59 | #592 |
| 2 | lapa-v0.1.2-instruct | 47.84 | #672 |
| 3 | MamayLM-Gemma-3-12B-IT-v1.0 | 47.14 | #584 |
| 4 | Claude Haiku 4.5 (20251001) | 46.99 | #261 |
| 5 | EuroLLM-22B-Instruct-2512 | 46.07 | #701 |
| 6 | Gemma 3 4B (IT) | 45.57 | #992 |
| 7 | Qwen 2.5 72B Instruct | 45.03 | #454 |
| 8 | Gemma 3 27B (IT) | 44.96 | #523 |
| 9 | Gemma 4 31B (IT) | 44.17 | #176 |
| 10 | Grok 4.1 Fast (Reasoning) | 43.54 | #217 (Grok 4.1 Fast) |
| 11 | Gemini 2.5 Flash Lite (Non-reasoning) | 43.5 | #422 (Gemini 2.5 Flash Lite) |
| 12 | Gemma 3n E4B (IT) | 43.48 | #749 |
| 13 | Granite 4.0 H Small | 43.29 | #665 |
| 14 | Llama 3.3 70B Instruct | 43.28 | #534 |
| 15 | Qwen 3 30B A3B 2507 Instruct | 43.02 | #488 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-croatian-nlu-mms-hr · How It Works · Data refreshed daily, snapshot 2026-10-20.