Open-Ended Question Answering on Earth Observation
97.05Judge ScoreQwen3
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen3Size (B)=235-A222026.03 | 97.05 | 48.24 | 2.17 | |
| GPT-4.1Size (B)=1800*2026.03 | 96.48 | 49.77 | 2.83 | |
| Mistral Medium 3.1Size (B)=200*2026.03 | 96.45 | 50.2 | 4.17 | |
| EVE-InstructSize (B)=242026.03 | 96.4 | — | 3.5 | |
| GPT OSSSize (B)=120A52026.03 | 94.2 | 50.3 | 4.83 | |
| GPT-5 nanoSize (B)=20*2026.03 | 92.2 | 50.2 | 5.33 | |
| MiniMax m2.5Size (B)=230A102026.03 | 91 | 51.1 | 5.17 |