Web Navigation on WebVoyager curated subset of 129 tasks
65.9Success Rate (SR)WebUncertainty
Evaluation Results
| Method | Links | |
|---|---|---|
| WebUncertaintyBackbone=GPT-4-Turbo2026.04 | 65.9 | |
| AgentOccamBackbone=GPT-4-Turbo2026.04 | 64.3 | |
| WebUncertaintyBackbone=Qwen-Max2026.04 | 63.6 | |
| WebPilotBackbone=GPT-4-Turbo2026.04 | 62 | |
| Agent-EBackbone=GPT-4-Turbo2026.04 | 59.7 | |
| AgentOccamBackbone=Qwen-Max2026.04 | 58.9 | |
| WebPilotBackbone=Qwen-Max2026.04 | 55.8 | |
| Agent-EBackbone=Qwen-Max2026.04 | 54.3 | |
| Browser UseBackbone=GPT-4-Turbo2026.04 | 51.9 | |
| WebVoyager-repBackbone=GPT-4-Turbo2026.04 | 50.4 | |
| Browser UseBackbone=Qwen-Max2026.04 | 48.8 | |
| WebVoyager-repBackbone=Qwen-Max2026.04 | 46.5 |