Terminal-based agentic execution on Terminal-Bench
53.8ScoreAEVO Procedure
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AEVO ProcedureCategory=Ours, Model=Gemini-3-Flash2026.05 | 53.8 | 7 | |
| DGMCategory=Procedure-Based Evolution, Model=Gemini-3-Flash2026.05 | 44.3 | 19 | |
| AFlowCategory=Procedure-Based Evolution, Model=Gemini-3-Flash2026.05 | 44.3 | 11 | |
| SPOCategory=Procedure-Based Evolution, Model=Gemini-3-Flash2026.05 | 42.9 | 19 | |
| GEPACategory=Procedure-Based Evolution, Model=Gemini-3-Flash2026.05 | 41.4 | 15 | |
| ADASCategory=Procedure-Based Evolution, Model=Gemini-3-Flash2026.05 | 38.6 | 7 | |
| ReAct (Pass@1)Category=Single-Agent Inference, Model=Gemini-3-Flash2026.05 | 28.6 | — |