Multi-Dimensional Reasoning Quality Evaluation on 975-item Reasoning Quality Evaluation Set benchmark leaderboard · SOTA2 Research

Multi-Dimensional Reasoning Quality Evaluation on 975-item Reasoning Quality Evaluation Set

0.872CQClaude-Haiku-4.5

Evaluation Results

Method	Links
Claude-Haiku-4.52026.05		0.872	0.417	0.963	0.85	0.643	0.926	0.778	0.797	0.761	0.752
DeepSeek-V32026.05		0.83	0.381	0.952	0.773	0.437	0.925	0.716	0.756	0.715	0.667
Gemini-2.5-Flash2026.05		0.808	0.372	0.952	0.819	0.493	0.918	0.727	0.755	0.711	0.682
LLaMA-3-70B2026.05		0.788	0.368	0.906	0.85	0.446	0.899	0.709	0.734	0.695	0.657
GPT-4o-mini2026.05		0.744	0.45	0.926	0.819	0.533	0.928	0.733	0.746	0.704	0.69
Phi-22026.05		0.495	0.432	0.608	0.869	0.406	0.828	0.606	0.566	0.546	0.547
Qwen2.5-1.5B2026.05		0.44	0.419	0.807	0.837	0.234	0.878	0.602	0.597	0.541	0.513