arXiv cs.AI / cs.LG / cs.CL·1d agoTokenCast: Forecasting Token Consumption During LLM Agent Execution#llm-agents#token-forecasting#cost-estimationAI research 2 sources