List your company
ResearchFormerly WizWand

Science Experimentation Reasoning on ScienceWorld (unseen)

88.6Success RateSKILLPYRAMID
49.0859.3469.679.86Jun 2, 2026Jun 7, 2026Jun 13, 2026Jun 19, 2026Jun 25, 2026Jul 1, 2026Jul 7, 2026
Updated 3mo ago

Evaluation results

MethodLinks
SKILLPYRAMIDModel=GPT-4.12026.0688.614.2
SKILLPYRAMIDModel=Qwen3-235B2026.0686.314.3
SKILLPYRAMIDModel=Gemini 2.5 Pro2026.0682.914.3
SKILLPYRAMIDModel=DeepSeek-V3.22026.0679.117
ReAct + SkillsModel=Gemini 2.5 Pro2026.0679.117
SkillRerankerBackbone Model=DeepSeek-v4-Flash2026.0774.4116.84
ExpeLModel=DeepSeek-V3.22026.0674.117.5
ReAct + SkillsModel=Qwen3-235B2026.067315.9
SkillRouterBackbone Model=DeepSeek-v4-Flash2026.0772.8518.14
LLM-as-selectorBackbone Model=DeepSeek-v4-Flash2026.0771.9519.59
ReAct + SkillsModel=GPT-4.12026.0671.818.6
Graph of SkillsBackbone Model=DeepSeek-v4-Flash2026.0770.6818.77
SkillRerankerBackbone Model=GPT-5.4-Mini2026.0769.6415.6
ReAct + SkillsModel=DeepSeek-V3.22026.0668.516.6
ReflexionModel=DeepSeek-V3.22026.0667.817.9
SkillRerankerBackbone Model=Qwen3.6-27B2026.0767.7313.08
ExpeLModel=Gemini 2.5 Pro2026.0667.414.9
SkillRouterBackbone Model=GPT-5.4-Mini2026.0766.9716.22
SkillRouterBackbone Model=Qwen3.6-27B2026.0766.8212.88
ExpeLModel=GPT-4.12026.0666.815.6
LLM-as-selectorBackbone Model=Qwen3.6-27B2026.0766.2413.19
LLM-as-selectorBackbone Model=GPT-5.4-Mini2026.0764.7217.18
ReActModel=DeepSeek-V3.22026.0664.719.3
ExpeLModel=Qwen3-235B2026.0663.816.2
Graph of SkillsBackbone Model=GPT-5.4-Mini2026.0763.4117.23
ReflexionModel=Gemini 2.5 Pro2026.0663.217.5
ReflexionModel=GPT-4.12026.0661.618.2
Graph of SkillsBackbone Model=Qwen3.6-27B2026.0761.0613.17
ReflexionModel=Qwen3-235B2026.0659.418.9
ReActModel=Gemini 2.5 Pro2026.0656.119.1
ReActModel=GPT-4.12026.0652.320.2
ReActModel=Qwen3-235B2026.0650.620.8