EVALITA - admission-test: leaderboard
The "admission-test" subtask of EVALITA. A task from EVALITA, the periodic evaluation campaign for Italian-language natural language processing and understanding.
Metric: CPS. Source: huggingface.co. Saturation forecast: Around January 2027. 48 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Mistral Large 2 (Nov) Instruct (2411) | 76.28 | #465 |
| 2 | Gemma 3 27B (IT) | 73.91 | #523 |
| 3 | Llama-3.1-Tulu-3-70B | 72.44 | #482 |
| 4 | Llama 3.3 70B Instruct | 70.72 | #534 |
| 5 | Mistral Small 3 | 69.77 | #627 |
| 6 | Qwen 2.5 72B Instruct | 69.3 | #454 |
| 7 | Gemma 2 27B (IT) | 68.47 | #645 |
| 8 | Llama 4 Scout Instruct | 68.17 | #539 |
| 9 | Gemma 3 12B (IT) | 65.9 | #664 |
| 10 | DeepSeek R1 Distill Llama 70B | 65.65 | #594 |
| 11 | calme-3.2-instruct-78B | 64.68 | #347 |
| 12 | Qwen 3 Next 80B A3B Instruct | 62.53 | #368 |
| 13 | Phi-4 | 61.91 | #706 |
| 14 | Qwen 2.5 VL 32B Instruct | 61.53 | #453 |
| 15 | Phi-3-medium-4k-instruct | 61.47 | #952 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=evalita-admission-test · How It Works · Data refreshed daily, snapshot 2026-10-20.