Web Browsing Research on BrowseComp (BC)
67.8Pass@3Claude Opus 4.5
Evaluation Results
| Method | Links | |
|---|---|---|
| Claude Opus 4.5Model Category=Frontier Proprietary Agents2026.05 | 67.8 | |
| Quest-35BModel Category=Open-Weight Agents, Parameters=35B, Evaluation Strategy=discard-all strategy2026.05 | 64.6 | |
| GPT-5Model Category=Frontier Proprietary Agents2026.05 | 59.9 | |
| Gemini 3 ProModel Category=Frontier Proprietary Agents2026.05 | 59.2 | |
| OpenAI-DRModel Category=Frontier Proprietary Agents2026.05 | 51.5 | |
| Quest-35BModel Category=Open-Weight Agents, Parameters=35B2026.05 | 45.5 | |
| Tongyi-DRModel Category=Open-Weight Agents2026.05 | 43.4 | |
| Quest-30BModel Category=Open-Weight Agents, Parameters=30B2026.05 | 37 | |
| WeDASModel=Miro-30B2026.03 | 35 | |
| MiroFlowModel=Miro-30B2026.03 | 33 | |
| WeDASModel=GPT-5-MINI2026.03 | 30 | |
| MiroFlowModel=GPT-5-MINI2026.03 | 29 | |
| OpenResearcherModel Category=Open-Weight Agents2026.05 | 26.3 | |
| WebSailorModel=WebSailor-72B2026.03 | 18.96 | |
| WebSailorModel=WebSailor-32B2026.03 | 16.05 |