BrowseComp browsecomp Leaderboard
Auto-discovered from papers reporting BrowseComp (Accuracy). · Metric: Accuracy (higher is better) · 🟢 Updated 2h ago
| # | Model | Accuracy | Paper |
|---|---|---|---|
| 1 | Seed 2.0 | 77.30 | link |
| 2 | GPT 5.4 | 72.10 | link |
| 3 | Claude Sonnet 4.6 | 69.30 | link |
| 4 | GLM 5.1 | 68.00 | link |
| 5 | Gemini 3.1 Pro | 67.00 | link |
| 6 | Kimi-K2.6 | 62.40 | link |
| 7 | DeepSeek V4 Pro | 61.40 | link |
| 8 | Kimi-K2.5 | 60.60 | link |
| 9 | MiniMax M2.5 | 60.40 | link |
| 10 | GLM 5.0 | 59.00 | link |
| 11 | DeepSeek v3.2 | 51.40 | link |
| 12 | DeepSearch-World: Self-Distillation for Deep Search Agents in a Verifiable Environment | 31.20 | link |