List your company

Long-horizon agent performance on BrowseComp-ZH

64.2Overall ScoreSAM
5.02420.38735.7551.113May 23, 2026

Evaluation results

MethodLinks
SAMBackbone=GLM-4.72026.0564.2
GLM-4.7CM=discard-tool2026.0562.5
Claude-4.5-OpusCM=w/o2026.0562.4
GLM-4.7CM=recent-k2026.0561.2
GLM-4.7CM=summary2026.0559
OpenAI-o3CM=w/o2026.0558.1
GLM-4.7CM=w/o2026.0552.5
AgentFold-30B-A3BCategory=Agent systems2026.0547.3
SAMBackbone=Qwen3.5-35B-A3B2026.0546.5
IterResearcher-30B-A3BCategory=Agent systems2026.0545.2
Qwen3.5-35B-A3BCM=recent-k2026.0545
Qwen3.5-35B-A3BCM=discard-tool2026.0543.5
Qwen3.5-35B-A3BCM=summary2026.0543
Qwen3.5-35B-A3BCM=w/o2026.0542.2
ReSum-30BCategory=Agent systems2026.0533.3
WebSailor-32BCategory=Agent systems2026.0525.5
WebThinker-32BCategory=Agent systems2026.057.3