Where the cost actually lives.

Every plan includes a monthly action allowance (Free 1K, Developer 10K, Starter 50K, Growth 500K, Enterprise 5M+). Beyond the allowance, each decision bills once in its highest applicable class: routine $0.002 (the fast policy-evaluation path, most actions), deliberation $0.05 (the multi-agent panel ran to completion), escalation $0.25 (a human reviewer was engaged). Shadow and test decisions are never metered, and the audit record is never metered.

Yes. Every GaaS plan includes deliberation inference — H2Om handles the LLM compute for multi-agent deliberation panels, always tracking the frontier model. There is a free tier (1K actions/month, verified email) and a Developer plan at $99/mo (10K actions). Paid tiers start at $500/mo Starter for 50K actions and 3 agents.

The Control Plane runs on GaaS infrastructure and handles governance logic, regulatory intelligence, policy evaluation, audit infrastructure, and learning systems — deterministic compute, and what your plan pays for. The Inference Plane handles LLM compute for deliberation agents and only triggers for high-risk, novel or conflicting actions. On GaaS Complete that inference is included; on GaaS Core (BYOM) the Inference Plane points at your own model endpoint instead.

Both run the identical governance pipeline — control plane, compliance knowledge base, audit trail, learning engine, and dashboard. Complete includes deliberation inference: we run the LLM compute for multi-agent panels. Core (BYOM) points deliberation at your own model endpoint, so you keep your provider, your rates, and your data residency — and the platform fee drops 30% ($350/mo Starter and $1,750/mo Growth, versus $500 and $2,500). Core starts at Starter; the Free and Developer plans are managed-inference only. On Core you also pay your own provider for deliberation tokens.

Beyond your plan’s included monthly actions, each decision bills once in its highest applicable class. On GaaS Complete: routine (approve/block/modify) $0.002, deliberation (multi-agent panel) $0.05, escalation (human review routing) $0.25. On GaaS Core (BYOM): routine $0.0015, deliberation $0.01 plus your own model tokens, escalation $0.25 — escalation is identical because human review involves no model compute. Fairness rules: shadow and test decisions are never metered, a failed deliberation bills as routine, and the audit record is never metered. Overage billing is active as of 1 August 2026, following a full report-only cycle; your included allowance is netted out first and is never metered, and usage recorded during the report-only period was not charged retroactively.

Pricing moves to per-decision usage billing scaled by action complexity and risk tier. Low-risk routine decisions are a fraction of a cent; high-stakes decisions with full deliberation are priced proportionately. You will never be charged without notice.

Yes, every governed action consumes one decision. You control which actions route through GaaS — only the consequential ones (payments, communications, record access). Most deployments govern a targeted subset, so your decision budget goes significantly further than it sounds.

No. Shadow Mode starts with just an email and a description of your agents. No credit card, no commitment, no contract.

Both use the same per-decision meter. Shadow Mode evaluates actions without enforcing decisions — it shows what GaaS would have done without affecting your operations. Activating enforcement uses the same cost structure, but decisions are actually applied.

Yes. Nonprofits, NGOs, and veteran-owned businesses qualify for free governance for life — contact us with your organization details. Early-stage startups building on GaaS are encouraged to reach out about extended access terms.