Article the Sixth · On the price of deliberation

On the price of deliberation.

Three plans, billed by credit. Begin without instrument; upgrade when the panel grows, the rounds lengthen, or the archive demands keeping.

Free
$0
Browse only
  • ·Explore a full sample deliberation
  • ·See every model answer + the reconciled memo
  • ·Subscribe to run your own questions
  • ·No card required
Recommended · Pro
$29/ mo
750 credits the month.
  • ·All 30+ models
  • ·Up to 12 models per panel
  • ·Up to 10 rounds
  • ·Expert personas
  • ·Web grounding
  • ·Personal MCP server
  • ·90-day archive retention
  • ·Credit top-ups ($0.06/credit)
Power
$79/ mo
2,500 credits the month.
  • ·Everything in Pro
  • ·Up to 30 models per panel
  • ·Deep runs fit comfortably
  • ·Final corpus analytics
  • ·API access
  • ·Peer-review mode
  • ·1-year archive retention
  • ·Credit top-ups ($0.05/credit)
For teams
Team
$49/ seat/mo
1,250 pooled / seat credits the month.
  • ·Everything in Power
  • ·3-seat minimum, pooled credits
  • ·Shared workspaces
  • ·Admin-managed MCP
  • ·Custom expert personas
  • ·Unlimited archive retention
  • ·Credit top-ups ($0.04/credit)
Business
$99/ seat/mo
Custom pooled
  • ·Everything in Team
  • ·SSO, SCIM, audit log
  • ·Custom model providers
  • ·Retention & data controls
  • ·Annual contract
Note · How the credit is reckoned

Of the credit, and what it represents.

§ I — The unit

A credit is the unit by which Plurality accounts for compute. One credit equals $0.01 of underlying model cost, marked up 2.5× to defray synthesis, deterministic analytics, web grounding, and the sundry overheads of the press. Per-query costs are rounded up to the nearest 0.1 credit, with a 0.5 credit minimum.

§ II — The formula
credits = max(
  0.5,
  ceil( (model_cost_usd / 0.01) × 2.5 × 10 ) / 10
)

model_cost_usd is the sum of input and output token costs for every model in the panel, summed across all rounds, plus a small synthesis fee per round.

§ III — Typical costs, by mode
  • Quick Take, Oracle, Forecast. Approximately 0.5 credits; single-round, cheap aggregation.
  • Consensus, Debate. Approximately 3–4 credits; five models, three rounds, mid-tier panel.
  • Red Team, Peer Review. Between 3–8 credits; depending on the depth of attack and the count of reviewers.

The frontier houses (Opus, GPT-5, Gemini 3) cost more than the middle tier; the query page renders the precise estimate before any submission.

§ IV — Of monthly credits, and top-ups

The plan's monthly credits reset at the start of each cycle and do not roll over. Paid plans may purchase top-up credits, which never expire and are spent only after the monthly allotment is exhausted. Top-ups are priced above the bundled rate — the monthly allotment is always the better deal, and higher plans get the better top-up rate.

§ V — On surprises, of which there are none

Every query renders an estimated credit range (min – max) against the present balance before submission. Insufficient balance is shown plainly; the run is blocked at the API layer if the balance does not cover the estimate.