Tool-use evaluation on OpenEarthAgent Step-by-step evaluation tool-agnostic rollouts benchmark leaderboard · SOTA2 Research

Tool-use evaluation on OpenEarthAgent Step-by-step evaluation tool-agnostic rollouts

0.9951Instruction AdherenceOpenEarthAgent

Evaluation Results

Method	Links
OpenEarthAgent2026.02		0.9951	0.9718	0.9608	0.621	0.8364
gpt-4o2026.02		0.9918	0.9388	0.8548	0.458	0.8676
Qwen3-Instruct-25072026.02		0.9734	0.8694	0.8412	0.3355	0.8328
Qwen2.5-Instruct2026.02		0.9408	0.8551	0.7846	0.38	0.8008
Qwen2.5-Instruct2026.02		0.8507	0.7213	0.6487	0.2412	0.6875
o4-mini2026.02		0.8368	0.6833	0.6417	0.3795	0.8948
Mistralv0.3-Instruct2026.02		0.6414	0.3512	0.2611	0.1245	0.2404
Llama-3.1-Instruct2026.02		0.4707	0.393	0.3411	0.1749	0.7908
Internlm3-Instruct2026.02		0.4454	0.3846	0.2782	0.1325	0.2984