EuroEval Croatian NLU: leaderboard
Croatian-language EuroEval natural-language-understanding suite, separating NLU task performance from the broader all-task EuroEval aggregate.
Metric: NLU Average Score (%). Source: euroeval.com. Saturation forecast: Around March 2027. 218 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 63.87 | #66 |
| 2 | Gemini 3.1 Flash Lite (Preview) | 60.27 | #194 |
| 3 | Claude Sonnet 4.6 | 59.87 | #90 |
| 4 | Gemini 2.5 Flash (Thinking) | 59.46 | #245 (Gemini 2.5 Flash) |
| 5 | Gemma 4 31B (IT) | 59.33 | #176 |
| 6 | GPT-5.4 Mini (2026-03-17) (High) | 59.28 | #169 (GPT-5.4 Mini (2026-03-17)) |
| 7 | GPT-5.2 | 58.97 | #110 |
| 8 | Claude Sonnet 4.5 (Thinking) | 58.88 | #142 (Claude Sonnet 4.5) |
| 9 | Gemini 2.5 Pro | 58.26 | #152 |
| 10 | Gemini 3 Flash (Preview) (Non-reasoning) | 58.15 | #82 (Gemini 3 Flash (Preview)) |
| 11 | Gemini 3 Flash (Preview) (Thinking) | 58.15 | #82 (Gemini 3 Flash (Preview)) |
| 12 | Qwen 3 32B (Non-reasoning) | 57.85 | #443 (Qwen 3 32B) |
| 13 | GPT-4.1 | 57.57 | #248 |
| 14 | Gemma 3 27B (IT) | 57.14 | #523 |
| 15 | Llama 3.1 70B | 56.68 | #592 |
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 · How It Works · Data refreshed daily, snapshot 2026-10-20.