Task Execution on TEXTCRAFT-SYNTH Hard (eval)
88Success RateRecursive Agent
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Recursive AgentContext Window (Train/Eval)=40K / 256K2026.05 | 88 | 694 | 73.3 | |
| Single AgentContext Window (Train/Eval)=40K / 256K2026.05 | 20 | 252 | 180 |
| Method | Links | |||
|---|---|---|---|---|
| Recursive AgentContext Window (Train/Eval)=40K / 256K2026.05 | 88 | 694 | 73.3 | |
| Single AgentContext Window (Train/Eval)=40K / 256K2026.05 | 20 | 252 | 180 |