EuroEval Romanian NLU - RONEC: leaderboard
The "RONEC" subtask of EuroEval Romanian NLU. Romanian-language EuroEval natural-language-understanding suite, separating NLU task performance from the broader all-task EuroEval aggregate.
Metric: Micro F1 excluding MISC (%). Source: euroeval.com. Saturation forecast: Around 2030. 207 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Mistral Small 3.1 | 52.34 | #610 |
| 2 | Mistral Small 3.2 | 50.91 | #605 |
| 3 | Voxtral-Small-24B-2507 | 49.78 | #581 |
| 4 | Gemma 3 12B (IT) | 48.7 | #664 |
| 5 | Gemma 4 31B (IT) | 48.5 | #176 |
| 6 | Llama 3.3 70B Instruct | 47.17 | #534 |
| 7 | Gemini 2.5 Flash (Thinking) | 47.06 | #245 (Gemini 2.5 Flash) |
| 8 | Gemma 3 27B (IT) | 46.68 | #523 |
| 9 | cogito-v1-preview-qwen-14B | 46.25 | |
| 10 | GPT-5 | 46.17 | #98 |
| 11 | Qwen 3 4B 2507 (Thinking) | 45.92 | #538 (Qwen 3 4B 2507) |
| 12 | gemma-3-27B-pt | 45.91 | #502 |
| 13 | Olmo-3.1-32B-Instruct-SFT | 45.89 | #879 |
| 14 | Claude Sonnet 4.5 (Thinking) | 45.63 | #142 (Claude Sonnet 4.5) |
| 15 | GPT-OSS-20B (Low) | 45.55 | #518 (GPT-OSS-20B) |
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-ronec · How It Works · Data refreshed daily, snapshot 2026-10-20.