EuroEval Italian NLU - ScaLA IT: leaderboard
The "ScaLA IT" subtask of EuroEval Italian NLU. Italian-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 March 2028. 435 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 80.08 | #66 |
| 2 | Gemini 3.1 Flash Lite (Preview) | 79.29 | #194 |
| 3 | Gemini 3 Flash (Preview) (Non-reasoning) | 77.58 | #82 (Gemini 3 Flash (Preview)) |
| 4 | Gemma 4 31B (IT) | 75.49 | #176 |
| 5 | Claude Sonnet 4.6 | 74.36 | #90 |
| 6 | Gemini 3 Flash (Preview) (Thinking) | 74.07 | #82 (Gemini 3 Flash (Preview)) |
| 7 | Qwen 3.5 35B A3B Base | 67.65 | #310 |
| 8 | GPT-5.2 | 66.85 | #110 |
| 9 | Gemma 4 26B A4B (IT) | 66.16 | #297 |
| 10 | Llama 3.1 70B | 64.33 | #592 |
| 11 | GPT-5.4 Mini (2026-03-17) (High) | 62.31 | #169 (GPT-5.4 Mini (2026-03-17)) |
| 12 | Qwen 3.5 9B | 62.14 | #379 |
| 13 | Grok 4.1 Fast (Reasoning) | 61.32 | #217 (Grok 4.1 Fast) |
| 14 | EuroLLM-22B-Instruct-2512 | 59.27 | #701 |
| 15 | NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 | 57.37 | #679 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-italian-nlu-scala-it · How It Works · Data refreshed daily, snapshot 2026-10-20.