{
  "schema_version": "newruntime-topic-hub-v0.1",
  "type": "topic_hub",
  "slug": "agent-economics",
  "title": "Agent economics - New Runtime",
  "description": "How AI work is priced, measured, routed, and justified by completed task value rather than raw token volume.",
  "answer": [
    "Agent economics moves measurement from token use to useful completed work.",
    "Cost only makes sense beside success rate, review time, retry loops, latency, and human attention.",
    "The mature metric is cost per accepted outcome, not cost per generation."
  ],
  "search_intents": [
    "agent economics",
    "AI coding cost",
    "cost per completed AI task"
  ],
  "status": "featured",
  "last_updated": "2026-07-22",
  "counts": {
    "total": 8,
    "signals": 6,
    "patterns": 1,
    "posts": 1,
    "atlas": 0,
    "sources": 9
  },
  "routes": {
    "html": "https://newruntime.com/topics/agent-economics/",
    "markdown": "https://newruntime.com/topics/agent-economics.md",
    "json": "https://newruntime.com/topics/agent-economics.json"
  },
  "top_sources": [
    "https://anthropic.com/news/claude-sonnet-5",
    "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-6-flash-3-5-flash-lite-3-5-flash-cyber/",
    "https://claude.com/blog/claude-model-and-effort-level-in-claude-code",
    "https://code.claude.com/docs/en/model-config",
    "https://cognition.com/blog/making-fable-cheaper-than-opus",
    "https://databricks.com/blog/benchmarking-coding-agents-databricks-multi-million-line-codebase",
    "https://engineering.ramp.com/post/ai-spend-value",
    "https://github.com/steipete/agent-scripts/blob/main/skills/codex-first/SKILL.md",
    "https://platform.claude.com/docs/en/agents-and-tools/tool-use/advisor-tool"
  ],
  "patterns": [
    {
      "kind": "Pattern",
      "slug": "agent-economics-moves-to-completed-work",
      "title": "Agent economics moves to completed work",
      "description": "The economically meaningful unit for agent systems is becoming cost per verified completed task rather than cost per token or model call.",
      "date": "2026-07-15",
      "topics": [
        "agent-economics",
        "model-routing",
        "product-metrics"
      ],
      "source_count": 5,
      "metric": "high",
      "url": "https://newruntime.com/patterns/agent-economics-moves-to-completed-work/"
    }
  ],
  "field_notes": [
    {
      "kind": "Field Note",
      "slug": "gemini-3-6-flash-agent-efficiency",
      "title": "Gemini 3.6 Flash Moves the Agent Race Toward Cost per Task",
      "description": "Google's Gemini 3.6 Flash release frames the model race around token efficiency, built-in computer use, and specialized cyber agents rather than raw chat intelligence alone.",
      "date": "2026-07-22",
      "topics": [
        "gemini",
        "agent-economics",
        "coding-agents"
      ],
      "source_count": 1,
      "url": "https://newruntime.com/posts/gemini-3-6-flash-agent-efficiency/"
    }
  ],
  "raw_signals": [
    {
      "kind": "Raw Signal",
      "slug": "fable-cost-per-completed-task",
      "title": "A pricier model can be cheaper per completed task",
      "description": "Cognition reports that Fable 5 completed coding work with fewer steps and output tokens than its previous lead model.",
      "date": "2026-07-15",
      "topics": [
        "agent-economics",
        "coding-agents",
        "model-routing"
      ],
      "source_count": 1,
      "metric": "structural",
      "url": "https://newruntime.com/signals/fable-cost-per-completed-task/"
    },
    {
      "kind": "Raw Signal",
      "slug": "claude-code-model-and-effort-are-separate-controls",
      "title": "Claude Code separates model choice from effort",
      "description": "Anthropic exposes model selection and effort level as different controls for capability, token use, latency, and persistence.",
      "date": "2026-07-13",
      "topics": [
        "coding-agents",
        "model-routing",
        "agent-economics"
      ],
      "source_count": 2,
      "metric": "notable",
      "url": "https://newruntime.com/signals/claude-code-model-and-effort-are-separate-controls/"
    },
    {
      "kind": "Raw Signal",
      "slug": "databricks-benchmarks-real-coding-agent-economics",
      "title": "Databricks benchmarks coding agents on its own codebase",
      "description": "Databricks evaluates agents on fresh internal pull-request tasks and measures success alongside runtime, tokens, and cost.",
      "date": "2026-07-12",
      "topics": [
        "coding-agents",
        "evals",
        "agent-economics"
      ],
      "source_count": 1,
      "metric": "structural",
      "url": "https://newruntime.com/signals/databricks-benchmarks-real-coding-agent-economics/"
    },
    {
      "kind": "Raw Signal",
      "slug": "advisor-model-guides-cheaper-executor",
      "title": "An advisor model can guide a cheaper executor",
      "description": "The advisor-tool pattern lets a fast executor request bounded analysis from a stronger model while keeping control of the task loop.",
      "date": "2026-07-09",
      "topics": [
        "model-routing",
        "orchestration",
        "agent-economics"
      ],
      "source_count": 2,
      "metric": "notable",
      "url": "https://newruntime.com/signals/advisor-model-guides-cheaper-executor/"
    },
    {
      "kind": "Raw Signal",
      "slug": "sonnet-moves-agent-capability-downmarket",
      "title": "Sonnet moves agent capability down the price curve",
      "description": "Anthropic positions Claude Sonnet 5 for planning, terminal work, browser use, and multi-step agent tasks at a lower tier.",
      "date": "2026-07-01",
      "topics": [
        "models",
        "agent-economics",
        "coding-agents"
      ],
      "source_count": 1,
      "metric": "notable",
      "url": "https://newruntime.com/signals/sonnet-moves-agent-capability-downmarket/"
    },
    {
      "kind": "Raw Signal",
      "slug": "ai-spend-measured-per-successful-task",
      "title": "AI spend should be measured per successful task",
      "description": "Ramp argues for allocating AI cost by use case, owner, completed outcome, failures, retries, review effort, and latency.",
      "date": "2026-06-27",
      "topics": [
        "agent-economics",
        "ai-strategy",
        "product-metrics"
      ],
      "source_count": 1,
      "metric": "structural",
      "url": "https://newruntime.com/signals/ai-spend-measured-per-successful-task/"
    }
  ],
  "atlas_records": []
}