Long-horizon agent performance on BrowseComp
60.6Overall ScoreKimi-K2.5-1T
Evaluation results
| Method | Links | |
|---|---|---|
| Kimi-K2.5-1TCategory=Foundation models, CM=w/o2026.05 | 60.6 | |
| SAMCategory=Models with CM, Backbone=GLM-4.72026.05 | 56.5 | |
| GPT-5.4Category=Foundation models, CM=w/o2026.05 | 54.9 | |
| GLM-4.7Category=Models with CM, CM=summary2026.05 | 53.5 | |
| GLM-4.7Category=Models with CM, CM=recent-k2026.05 | 51.5 | |
| OpenAI-o3Category=Foundation models, CM=w/o2026.05 | 49.7 | |
| GLM-4.7Category=Models with CM, CM=discard-tool2026.05 | 49 | |
| GLM-4.7Category=Foundation models, CM=w/o2026.05 | 43.5 | |
| SAMCategory=Models with CM, Backbone=Qwen3.5-35B-A3B2026.05 | 42.2 | |
| Qwen3.5-35B-A3BCategory=Models with CM, CM=discard-tool2026.05 | 40.5 | |
| Qwen3.5-35B-A3BCategory=Models with CM, CM=summary2026.05 | 39.5 | |
| Qwen3.5-35B-A3BCategory=Models with CM, CM=recent-k2026.05 | 38.2 | |
| IterResearcher-30B-A3BCategory=Agent systems2026.05 | 37.3 | |
| Claude-4.5-OpusCategory=Foundation models, CM=w/o2026.05 | 37 | |
| AgentFold-30B-A3BCategory=Agent systems2026.05 | 36.2 | |
| Qwen3.5-35B-A3BCategory=Foundation models, CM=w/o2026.05 | 36 | |
| ReSum-30BCategory=Agent systems2026.05 | 18.3 | |
| WebSailor-32BCategory=Agent systems2026.05 | 10.5 | |
| WebThinker-32BCategory=Agent systems2026.05 | 2.8 |