Multiple Choice Question Answering (Multiple) on Earth Observation
87.56IoUGPT-4.1
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GPT-4.1Size (B)=1800*2026.03 | 87.56 | 78.19 | 2.83 | |
| Qwen3Size (B)=235-A222026.03 | 87.4 | 80.97 | 2.17 | |
| EVE-InstructSize (B)=242026.03 | 86.12 | 77.73 | 3.5 | |
| Mistral Medium 3.1Size (B)=200*2026.03 | 85.44 | 76.33 | 4.17 | |
| MiniMax m2.5Size (B)=230A102026.03 | 84.82 | 77.72 | 5.17 | |
| GPT OSSSize (B)=120A52026.03 | 84.56 | 76.79 | 4.83 | |
| GPT-5 nanoSize (B)=20*2026.03 | 84.4 | 76.1 | 5.33 |