---
schema_version: "newruntime-agent-readable-v0.1"
type: "trend_pattern"
slug: "agent-economics-moves-to-completed-work"
title: "Agent economics moves to completed work"
description: "Token prices and subscription fees are weak proxies when turns, retries, delegation, verification, latency, and successful outcomes drive the real cost."
thesis: "The economically meaningful unit for agent systems is becoming cost per verified completed task rather than cost per token or model call."
status: "published"
confidence: "high"
first_seen: "2026-06-27"
last_verified: "2026-07-15"
supporting_signals: ["tg-2696","tg-2668","tg-2676","tg-2646","tg-2545"]
related_posts: ["coding-agent-cost-environment-config","ridge-ai-spend-process-cost","model-match-effort-selection"]
counter_evidence: ["Token accounting remains necessary for capacity planning and can approximate cost in simple, stable, single-call workflows.","Task-level attribution can be expensive or ambiguous when one agent run contributes to several downstream outcomes."]
revision_trigger: "Revise the thesis if token-level optimization reliably predicts lower total task cost without measuring failures, retries, human review, or completion quality."
topics: ["agent-economics","model-routing","product-metrics"]
source_urls: ["https://cognition.com/blog/making-fable-cheaper-than-opus","https://databricks.com/blog/benchmarking-coding-agents-databricks-multi-million-line-codebase","https://claude.com/blog/claude-model-and-effort-level-in-claude-code","https://platform.claude.com/docs/en/agents-and-tools/tool-use/advisor-tool","https://engineering.ramp.com/post/ai-spend-value"]
routes: {"html":"https://newruntime.com/patterns/agent-economics-moves-to-completed-work/","markdown":"https://newruntime.com/patterns/agent-economics-moves-to-completed-work.md","json":"https://newruntime.com/patterns/agent-economics-moves-to-completed-work.json"}
source_format: "markdown"
---

# Agent economics moves to completed work

## Why token price is incomplete

An agent run accumulates cost through repeated context, tool calls, failed
attempts, delegated work, and human verification. A more expensive lead model
can lower total cost when it takes fewer turns, delegates earlier, writes a
better task contract, and avoids redoing sidekick work.

## Supporting evidence

- Cognition reports that Fable plus the same sidekick cost less per benchmark
  run than Opus plus sidekick despite a higher token price. The observed
  difference came from turn count, context use, and delegation behavior.
- Databricks measures coding-agent outcomes against internal tasks together with
  time, tokens, and cost rather than relying on a public benchmark score alone.
- Claude Code exposes effort as a separate runtime control from the model,
  making cost and persistence tunable per task.
- Ramp argues for assigning spend to use cases and completed outcomes, including
  retries and human review.

## Operational consequence

Each run should record task identity, route, models, turns, tool calls, elapsed
time, retries, review effort, outcome, and failure class. Cost optimization can
then target the whole trajectory instead of a single line item.