Language Understanding on ITALIC
86.6AccuracyGPT-5 nano
Evaluation results
| Method | Links | |
|---|---|---|
| GPT-5 nanoScale=nano2026.05 | 86.6 | |
| Ministral-3-8BParameters=8B2026.05 | 82.5 | |
| gemma-3-12b-itParameters=12B2026.05 | 77.1 | |
| Qwen3-8BParameters=8B2026.05 | 73 | |
| Llama-3.1-8BParameters=8B2026.05 | 68.3 | |
| Qwen3-4BParameters=4B2026.05 | 67.8 | |
| Velvet-14BParameters=14B2026.05 | 66.6 | |
| gemma-3-4b-itParameters=4B2026.05 | 63.4 | |
| LLaMAntino-3-8BParameters=8B2026.05 | 62.5 | |
| EngGPT2-16B-A3BParameters=16B2026.05 | 60.9 | |
| gpt-oss-20bParameters=20B2026.05 | 58.9 | |
| Moonlight-16B-A3BParameters=16B2026.05 | 58.2 | |
| Llama-3.2-3BParameters=3B2026.05 | 55.7 | |
| deepseek-moe-16bParameters=16B2026.05 | 46.6 | |
| FastwebMIIA-7BParameters=7B2026.05 | 43.7 | |
| Minerva-7BParameters=7B2026.05 | 40.6 |