List your company

Open-Ended Question Answering (with Context) on Earth Observation

86.65Judge ScoreGPT-4.1
77.945280.205182.46584.7249Mar 20, 2026

Evaluation results

MethodLinks
GPT-4.1Size (B)=1800*2026.0386.6549.222.83
Mistral Medium 3.1Size (B)=200*2026.0386.4450.994.17
Qwen3Size (B)=235-A222026.0386.150.092.17
GPT OSSSize (B)=120A52026.0384.850.74.83
GPT-5 nanoSize (B)=20*2026.0384.448.65.33
MiniMax m2.5Size (B)=230A102026.0381.5751.25.17
EVE-InstructSize (B)=242026.0378.28—3.5