Spend transparency

OpenGTM shows you the bill before a run and keeps a ledger of what every provider actually cost. None of this is a premium tier.

The estimate

GET /api/workbooks/{id}/run/estimate counts rows, resolves the provider chain for every enrichment column, and returns:

{
  "rows": 50,
  "worst_usd": 6.00,
  "best_usd": 1.00,
  "breakdown": [
    { "column": "Email", "paid_providers": ["hunter_io", "apollo_io", "snovio", "prospeo"],
      "worst_usd": 6.00, "best_usd": 1.00 }
  ],
  "note": "worst = every paid provider tried per row; best = cheapest paid hit first. Cross-provider cache + confidence early-exit reduce actual spend."
}

Worst case multiplies rows by the sum of every paid provider in the chain; best case uses the cheapest paid provider only. Free providers are not priced. The UI gates the run behind a confirmation using these numbers.

Provider prices

The vendor catalog carries a list price per lookup for each paid provider (for example Hunter $0.04, Apollo $0.03, Snov.io $0.03, Prospeo $0.02, People Data Labs $0.03, Abstract $0.01, DeBounce $0.008, NumVerify $0.005, Google Maps $0.005, LeadMagic $0.05, Prospeo mobile $0.10). A provider’s own declared cost_per_lookup, including those in YAML manifests, overrides the table. Research columns are priced at $0.05 per cell only when RESEARCH_VENDOR_COST=1, so enabling it never retroactively changes an existing estimate.

Search-style operations scale by page count and bulk operations by record count.

The ceiling

PUT /api/workbooks/{id}/budget with {"max_usd": 5} sets a spend ceiling (0 means unlimited). GET .../cost returns the ceiling, the amount spent, and the remainder. Once the remainder is smaller than a provider’s price, the planner drops that provider from every chain. Agent columns additionally respect their own policy.max_cost_usd and never exceed the workbook’s headroom.

The ledger

Every attempt records cost, latency, hit or miss, and confidence per provider and field. Read it at GET /api/workbooks/meta/provider-stats and .../provider-accuracy, or per cell through the trace endpoint. The same data drives the planner.

Billing is optional

BILLING_ENABLED is off by default. Off means runs are never blocked and no debits happen, which is the right mode for a self-hosted install. On, a workspace credit ledger, an idempotent debit, a 402 gate and a Stripe top-up webhook exist under /api/billing. Platform-metered costs such as outreach sends and poller runs default to $0.00.