Tool Use on BrowseComp Domain-specific (9) Search
22.5AccuracyGold Oracle
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Gold OracleModel=GPT-5, Strategy=ReAct2026.02 | 22.5 | 7 | |
| TOOLOBSERVERModel=GPT-5, Strategy=ReAct2026.02 | 21.9 | 6.3 | |
| EasyToolModel=GPT-5, Strategy=ReAct2026.02 | 18.9 | 8.2 | |
| Base ReActModel=GPT-5, Strategy=ReAct2026.02 | 18.1 | 10 | |
| Play2PromptModel=GPT-5, Strategy=ReAct2026.02 | 17.9 | 7.8 | |
| Gold OracleModel=GPT-5-mini, Strategy=ReAct2026.02 | 9.7 | 2.8 | |
| EasyToolModel=GPT-5-mini, Strategy=ReAct2026.02 | 7.1 | 3.8 | |
| TOOLOBSERVERModel=GPT-5-mini, Strategy=ReAct2026.02 | 3.2 | 3.5 | |
| Base ReActModel=GPT-5-mini, Strategy=ReAct2026.02 | 3.1 | 4.1 | |
| Play2PromptModel=GPT-5-mini, Strategy=ReAct2026.02 | 2.8 | 3.5 |