EuroEval Romanian NLU - MultiWikiQA RO: leaderboard
The "MultiWikiQA RO" subtask of EuroEval Romanian NLU. Romanian-language EuroEval natural-language-understanding suite, separating NLU task performance from the broader all-task EuroEval aggregate.
Metric: Character F1 (%). Source: euroeval.com. Saturation forecast: Not forecast. 207 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | SOLAR-10.7B-v1.0 | 77.57 | #1253 |
| 2 | Bielik-11B-v2.3-Instruct | 76.93 | #693 |
| 3 | Mistral Small 3.2 | 76.89 | #605 |
| 4 | gemma-3-27B-pt | 76.12 | #502 |
| 5 | Qwen 3 32B (Non-reasoning) | 75.86 | #443 (Qwen 3 32B) |
| 6 | Llama 3.1 70B | 75.64 | #592 |
| 7 | Qwen 3 14B (Non-reasoning) | 75.44 | #537 (Qwen 3 14B) |
| 8 | Olmo-3.1-32B-Instruct-SFT | 75.08 | #879 |
| 9 | Voxtral-Small-24B-2507 | 74.16 | #581 |
| 10 | Qwen3.5-4B-Base | 74.08 | #798 |
| 11 | Olmo-3.1-32B-Instruct-DPO | 73.9 | #845 |
| 12 | gemma-3-12B-pt | 73.8 | #686 |
| 13 | Mistral Small 3.1 | 73.72 | #610 |
| 14 | Ministral-3-14B-Reasoning-2512 | 72.98 | #540 |
| 15 | Llama 3.1 8B | 72.78 | #1151 |
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-multiwikiqa-ro · How It Works · Data refreshed daily, snapshot 2026-10-20.