EuroEval Romanian NLU: leaderboard
Romanian-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: Estimated already saturated. 212 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 69.29 | #66 |
| 2 | Qwen 3 32B (Non-reasoning) | 68.83 | #443 (Qwen 3 32B) |
| 3 | Claude Sonnet 4.6 | 67.28 | #90 |
| 4 | Mistral Small 3.1 | 67.2 | #610 |
| 5 | Gemma 4 31B (IT) | 67.17 | #176 |
| 6 | Gemini 2.5 Flash (Thinking) | 67.16 | #245 (Gemini 2.5 Flash) |
| 7 | GPT-5.4 Mini (2026-03-17) (High) | 67.06 | #169 (GPT-5.4 Mini (2026-03-17)) |
| 8 | Mistral Small 3.2 | 67.01 | #605 |
| 9 | Gemma 3 27B (IT) | 66.86 | #523 |
| 10 | Gemini 3 Flash (Preview) (Non-reasoning) | 66.66 | #82 (Gemini 3 Flash (Preview)) |
| 11 | Qwen 3 14B (Non-reasoning) | 66.46 | #537 (Qwen 3 14B) |
| 12 | gemma-3-27B-pt | 66.29 | #502 |
| 13 | GPT-5 | 66.27 | #98 |
| 14 | GPT-5.2 | 66.26 | #110 |
| 15 | Gemini 3 Flash (Preview) (Thinking) | 66.16 | #82 (Gemini 3 Flash (Preview)) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-romanian-nlu · How It Works · Data refreshed daily, snapshot 2026-10-20.