← Home · Quickstart · Router · Agents · Docs · Dashboard

Pricing model

How pricing is calculated

Public run pricing is made of two parts: the selected lane’s base price plus a fixed 0.005 USDC orchestration fee on POST /api/run.

run_price_usdc = base_agent_price_usdc + orchestration_fee_usdc

orchestration_fee_usdc = 0.005

examples:
comedian-v1   = 0.005 + 0.005 = 0.010
writer-v1     = 0.010 + 0.005 = 0.015
support-v1    = 0.010 + 0.005 = 0.015
analyst-v1    = 0.015 + 0.005 = 0.020
coder-v1      = 0.025 + 0.005 = 0.030
strategist-v1 = 0.025 + 0.005 = 0.030
researcher-v2 = 0.025 + 0.005 = 0.030

Free discovery endpoints

These endpoints help builders understand cost and fit before they commit spend.

• GET /api/health — liveness and dependency check.
• GET /api/capabilities — machine-readable capability manifest.
• GET /api/pricing — public pricing and cost discovery.
• GET /api/providers — provider and route discovery.
• POST /api/route — task-to-agent recommendation before execution.

GET https://attestify-os.vercel.app/api/pricing

{
  "currency": "USDC",
  "network": "eip155:8453",
  "pricing_version": "2026-05-tiered-v1",
  "cost_model": "2026-05-cost-v1",
  "orchestration_fee_usdc": 0.005,
  "cheap_test": {
    "agent_id": "comedian-v1",
    "base_agent_price_usdc": 0.005,
    "run_price_usdc": 0.010
  }
}

Run pricing by lane

These lanes line up with the current tiered pricing model used by POST /api/run.

Paid execution endpoints

Paid endpoints perform real work and return evidence around that work.

Full pricing matrix

Request patterns

The pricing model works best when paired with the router-first request shape. Use steering when you want soft guidance, and explicit overrides only when you need a hard pin.

{
  "session_id": "pricing-demo-001",
  "task": "Tell me a short joke about AI agents",
  "preferred_agent_id": "comedian-v1",
  "options": {
    "include_memory": true,
    "write_memory": true,
    "verify": true
  }
}

{
  "session_id": "pricing-demo-002",
  "task": "Refactor this API route to return structured errors and add receipt logging",
  "agent_id": "coder-v1",
  "options": {
    "include_memory": true,
    "write_memory": true,
    "verify": true
  }
}

Route recommendation

When the right lane is unclear, call POST /api/route before the paid run. This is a free helper for recommendation, not the paid work unit itself.

POST https://attestify-os.vercel.app/api/route
Content-Type: application/json

{
  "task": "I need help deciding whether this request belongs to an analyst, strategist, or researcher lane."
}

After you inspect the recommendation, you can either send the paid request with preferred_agent_id to steer execution, or set agent_id to pin the selected lane.

Why this model works

• Builders can inspect what exists before spending.
• Agents and developers can estimate cost from machine-readable pricing.
• Cheap smoke tests make end-to-end proofs inexpensive.
• Premium routes make it easy to pay more only when the task justifies it.
• Attestify charges when it performs durable paid work and returns evidence.

Control Tower pricing

Control Tower is separate from OS run pricing. It is the governance layer for approvals, spend controls, policy, allowlists, and stronger evidence around sensitive workflows.

Early Control Tower pricing will likely be bespoke for design partners, then evolve into packaging based on governed environments, workflow volume, or managed policy scope.

Next steps

• Start with the Quickstart.
• Read the Router page to understand route selection and overrides.
• Browse Agents for fit, examples, and when to use each lane.
• Read the Docs for request and response shapes.
• Use the Dashboard to inspect public execution activity.