BrowseComp
Search · 2025-04-10
Hard web-browsing benchmark for locating obscure facts.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Mythos 5 | 93.3 |
| GPT-5.6 Sol | 92.2 |
| Kimi K3 | 91.2 |
| Claude Opus 5 | 90.8 |
| GPT-5.5 Pro | 90.1 |
| GPT-5.4 Pro | 89.3 |
| Opus 4.8 | 88.5 |
| Claude Fable 5 | 88.0 |
| Claude Mythos Preview | 87.9 |
| GPT-5.6 Terra | 87.5 |
| Gemini 3.1 Pro Preview | 85.9 |
| Sonnet 5 | 84.7 |
| GPT-5.5 | 84.4 |
| GPT-5.6 Luna | 84.0 |
| Opus 4.6 | 84.0 |
| DeepSeek-V4-Pro | 83.4 |
| Kimi K2.6 | 83.2 |
| GPT-5.4 | 82.7 |
| Opus 4.7 | 79.3 |
| Qwen3.5 397B A17B | 78.6 |
| GPT-5.2 Pro | 77.9 |
| Inkling-Small | 77.4 |
| GPT-5.3-Codex | 77.3 |
| Inkling | 77.1 |
| MiniMax M2.7 | 76.3 |
| Kimi K2.5 | 74.9 |
| Sonnet 4.6 | 74.7 |
| DeepSeek-V4-Flash | 73.2 |
| GPT-5.2 | 65.8 |
| Nemotron 3 Ultra 550B A55B | 63.0 |
| Gemini 3 Pro Preview | 59.2 |
| O1 | 9.9 |
| GPT-4o Search Preview | 1.9 |
| GPT-4.5 | 0.9 |
| GPT-4o (2024-08-06) | 0.6 |