List your company

Agentic Task Success on Baba Is AI

38Success RateDynamic Cheatsheet
8.3616.05523.7531.445Feb 8, 2026

Evaluation results

MethodLinks
Dynamic CheatsheetFoundation Model=GPT-5-mini2026.0238
ALMAFoundation Model=GPT-5-mini2026.0233.3
Trajectory RetrievalFoundation Model=GPT-5-mini2026.0230.9
G-MemoryFoundation Model=GPT-5-mini2026.0226.2
No MemoryFoundation Model=GPT-5-mini2026.0221.4
Reasoning BankFoundation Model=GPT-5-mini2026.0221.4
Trajectory RetrievalFoundation Model=GPT-5-nano2026.0219
ALMAFoundation Model=GPT-5-nano2026.0219
G-MemoryFoundation Model=GPT-5-nano2026.0214.3
No MemoryFoundation Model=GPT-5-nano2026.029.5
Reasoning BankFoundation Model=GPT-5-nano2026.029.5
Dynamic CheatsheetFoundation Model=GPT-5-nano2026.029.5