List your company

Agentic Task Success on ALFWorld

87.1Success RateALMA
-0.46822.2664567.734Feb 8, 2026

Evaluation results

MethodLinks
ALMAFoundation Model=GPT-5-mini2026.0287.1
Trajectory RetrievalFoundation Model=GPT-5-mini2026.0280
Dynamic CheatsheetFoundation Model=GPT-5-mini2026.0278.6
G-MemoryFoundation Model=GPT-5-mini2026.0274.8
No MemoryFoundation Model=GPT-5-mini2026.0267.6
Reasoning BankFoundation Model=GPT-5-mini2026.0267.1
ALMAFoundation Model=GPT-5-nano2026.0212.4
G-MemoryFoundation Model=GPT-5-nano2026.027.6
Dynamic CheatsheetFoundation Model=GPT-5-nano2026.025.7
Trajectory RetrievalFoundation Model=GPT-5-nano2026.025.2
Reasoning BankFoundation Model=GPT-5-nano2026.025.2
No MemoryFoundation Model=GPT-5-nano2026.022.9