Agentic Task Success on ALFWorld
87.1Success RateALMA
Evaluation Results
| Method | Links | |
|---|---|---|
| ALMAFoundation Model=GPT-5-mini2026.02 | 87.1 | |
| Trajectory RetrievalFoundation Model=GPT-5-mini2026.02 | 80 | |
| Dynamic CheatsheetFoundation Model=GPT-5-mini2026.02 | 78.6 | |
| G-MemoryFoundation Model=GPT-5-mini2026.02 | 74.8 | |
| No MemoryFoundation Model=GPT-5-mini2026.02 | 67.6 | |
| Reasoning BankFoundation Model=GPT-5-mini2026.02 | 67.1 | |
| ALMAFoundation Model=GPT-5-nano2026.02 | 12.4 | |
| G-MemoryFoundation Model=GPT-5-nano2026.02 | 7.6 | |
| Dynamic CheatsheetFoundation Model=GPT-5-nano2026.02 | 5.7 | |
| Trajectory RetrievalFoundation Model=GPT-5-nano2026.02 | 5.2 | |
| Reasoning BankFoundation Model=GPT-5-nano2026.02 | 5.2 | |
| No MemoryFoundation Model=GPT-5-nano2026.02 | 2.9 |