ResearchBenchmarksEmbodied Task Planning on ALFWorld standard evaluation set (134 tasks)Follow97Pass@1 AccuracyGiG59.5669.287988.72Jan 29, 2026Evaluation ResultsMethodMethodLinksPass@1 AccuracyGiGModel=Qwen3Model=Qwen32026.0197GiGModel=DeepSeekModel=DeepSeek2026.0197GiGModel=GeminiModel=Gemini2026.0191ReCAPModel=Qwen3Model=Qwen32026.0189ReCAPModel=GeminiModel=Gemini2026.0186ReCAPModel=DeepSeekModel=DeepSeek2026.0182ReActModel=Qwen3Model=Qwen32026.0161