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

Embodied Agent Instruction Following on ALFWorld (ALF) (test)

91Success Rate

EPPO

Updated 17d ago

Evaluation Results

Method	Links
EPPO 2026.07		91
AgentRL 2026.07		89.2
AgentLM-70B 2026.07		86
AgentLM-7B 2026.07		84
BAPO 2026.07		80.9
AgentLM-13B 2026.07		76
Claude-Sonnet-4 2026.07		73.6
Claude-Sonnet-4 Thinking 2026.07		69
GPT-5 2026.07		65.4
DeepSeek-R1 2026.07		51.4
Hephaestus-8B-IFT 2026.07		46
o4-mini 2026.07		32.6
Qwen2.5-32B-Instruct 2026.07		32.1
DeepSeek-V3 2026.07		31.9
Hephaestus-8B-Base 2026.07		30
o3-mini 2026.07		28.4
Qwen2.5-14B-Instruct 2026.07		8.7
Qwen2.5-3B-Instruct 2026.07		2.2
DCPO 2026.07		0