Natural Language Understanding on italic cot custom
86.8Flexible ExtractGPT-5 nano
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5 nanostrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 86.8 | |
| Ministral-3-8B-Instruct-2512-BF16strategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 81.3 | |
| gemma-3-12b-itstrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 79 | |
| Qwen3-8Bstrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 77.4 | |
| gpt-oss-20bstrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 76.9 | |
| Qwen3-4Bstrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 71.4 | |
| EngGPT2-16B-A3Bstrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 68.3 | |
| LLaMAntino-3-ANITA-8B-Inst-DPO-ITAstrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 67.3 | |
| Velvet-14Bstrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 66.6 | |
| gemma-3-4b-itstrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 65.6 | |
| Llama-3.1-8B-Instructstrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 65.3 | |
| FastwebMIIA-7Bstrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 53.7 | |
| Moonlight-16B-A3B-Instructstrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 53 | |
| Llama-3.2-3B-Instructstrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 49.6 | |
| deepseek-moe-16b-chatstrategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 36.8 | |
| Minerva-7B-instruct-v1.0strategy=Chain-of-Thought, configuration=custom/structural improvements2026.05 | 32.9 |