List your company

Agentic Execution Trace Analysis on TraceBench Deep Research BrowseComp

81Overall ScoreTraceSIR
53.9660.986875.02Feb 28, 2026

Evaluation results

MethodLinks
TraceSIRBackbone=Claude-4.62026.02818.58.5878.51
ClaudeCodeBackbone=Claude-4.62026.02666.57.55.576.52
TraceSIRBackbone=GLM-52026.02657.56.55.5673
ClaudeCodeBackbone=GLM-52026.02555.54.55.5664