List your company

Trajectory Attribution on Who&When n=58 (Hand-Crafted)

73Agent-level AccuracyFALAT
36.49645.97355.4564.927May 30, 2026

Evaluation results

MethodLinks
FALATBackbone=DeepSeek V3.22026.057329.1
FALATBackbone=MiniMax 2.52026.0566.325.9
FALATBackbone=GPT-5.4-Mini2026.0565.522.2
AgentTracerBackbone=AgentTracer-8B2026.0563.820.7
DirectBackbone=Qwen3-Coder2026.0560.413.8
CHIEFBackbone=GPT-5.4-Mini2026.0555.213.8
DirectBackbone=DeepSeek-R12026.0553.46.9
DirectBackbone=Gemini-2.5-Pro2026.0551.76.9
CHIEFBackbone=MiniMax 2.52026.055117
CHIEFBackbone=DeepSeek V3.22026.055010.3
DirectBackbone=LLaMA-3.2-3B2026.05503.5
DirectBackbone=Claude-Sonnet-42026.055018.9
DirectBackbone=Qwen3-32B2026.0544.81.7
DirectBackbone=Qwen3-8B2026.0539.53.5
DirectBackbone=GPT-4.12026.0537.93.4