List your company

Agentic Task Success on Textworld

75Success RateALMA
-0.81618.86738.5558.233Feb 8, 2026

Evaluation results

MethodLinks
ALMAFoundation Model=GPT-5-mini2026.0275
G-MemoryFoundation Model=GPT-5-mini2026.0268.8
Trajectory RetrievalFoundation Model=GPT-5-mini2026.0267
No MemoryFoundation Model=GPT-5-mini2026.0260.5
Dynamic CheatsheetFoundation Model=GPT-5-mini2026.0257.8
Reasoning BankFoundation Model=GPT-5-mini2026.0256.1
ALMAFoundation Model=GPT-5-nano2026.026.2
No MemoryFoundation Model=GPT-5-nano2026.025.4
Reasoning BankFoundation Model=GPT-5-nano2026.025.3
Dynamic CheatsheetFoundation Model=GPT-5-nano2026.024.3
Trajectory RetrievalFoundation Model=GPT-5-nano2026.022.7
G-MemoryFoundation Model=GPT-5-nano2026.022.1