Share your thoughts, 1 month free Claude Pro on usSee more

Sokoban on Sokoban (Hard-1)

91.41Accuracy

GPT-5

Updated 3mo ago

Evaluation Results

Method	Links
GPT-5 2025.11		91.41
OpenAI o3 2025.11		90.23
OpenAI o4-mini 2025.11		69.14
WMAct 2025.11		52.68
PPO - Interactive 2025.11		41.26
Claude 4.5 Sonnet 2025.11		34.77
Gemini 2.5 Pro 2025.11		17.19
Qwen3-14B 2025.11		6.64
Qwen3-8B 2025.11		4.6
PPO - EntirePlan 2025.11		2.34
Qwen3-8B-Own 2025.11		0.84
GPT-4o 2025.11		0
Qwen2.5-7B-Instruct 2025.11		0
Qwen2.5-32B-Instruct 2025.11		0