Free
$0to start
The whole platform, free. Pay only for what you run.
Get started- 500 runtime-minutes free / month
- Up to 3 members
- Build, run + manage workflows
- Roles, org secrets, audit log
- Live run tails + full history
- Bring your own model key
Pricing
Free to start. Pay for what runs.
The whole platform is free — build, run, and govern every workflow at no cost. Everything that actually runs draws from a prepaid credit balance: the runtime a run holds a machine, plus managed model cost. Bring your own key and tokens are free from us. No synthetic credits, no surprise overage.
Team
$20per seat / month
For teams past 3 members: add seats, share one prepaid usage wallet.
Get started- Everything in Free
- Unlimited members
- Shared prepaid credit wallet
- Cron, webhook + workflow triggers
- Self-host the open-source runtime
- Top up the wallet from $5
Enterprise
Customcommitted use
Dedicated capacity and governance for orgs running at scale.
Contact sales- Everything in Team
- SSO / SAML + SCIM
- Dedicated capacity + SLA
- Committed-use credit discounts
- Data-retention controls
Read-only viewers are always free — share a run with a link.
How pricing works
Pay for what runs, nothing for access
Most agent platforms wrap your bill in a synthetic credit so you can't tell access from compute. Boardwalk keeps it simple — access is free, usage is prepaid credits — and lets you bring your own model or hardware to shrink the bill.
01
Access is free
For up to 3 members, building, running, and governing workflows costs nothing: roles, org secrets, an audit log, live tails, and full run history are all included. Past 3, seats are $20 a month; either way you only pay for usage on top.
02
Usage is prepaid credits
Every run draws from a prepaid balance: the runtime it holds a machine, plus managed model cost. Your balance is a hard ceiling, so there's no surprise overage — you top up what you intend to spend.
03
Bring your own, pay less
Route to your own model key and Boardwalk charges $0 on tokens — you settle with the provider directly. Or self-host the open-source runtime and run the same workflow files on your own hardware.
Runtime rates
Per active minute a run holds a machine, drawn from your credit balance (including time spent waiting on a sleep or a child run), and nothing when no run is executing. The rate scales with machine size; specialized (GPU, Windows, macOS) machines are on the way.
| Machine size | Shape | Rate |
|---|---|---|
| Small | 1 vCPU · 2 GB | $0.05 / min |
| Standard | 2 vCPU · 4 GB | $0.10 / min |
| Large | 4 vCPU · 8 GB | $0.20 / min |
| XL | 8 vCPU · 16 GB | $0.40 / min |
Model inference
BYO key = $0On the managed lane you pay the model's current rate plus a 5% margin, metered as real cost. Boardwalk routes to the cheapest available source and passes the lower price through. Bring your own key and tokens cost $0 from us; you settle with the provider directly.
See all models and rates →| Model | Input / Mtok | Output / Mtok |
|---|---|---|
| GPT-5.5 | $5.25 | $31.50 |
| Gemini 3 Pro | $2.10 | $12.60 |
| Claude Opus 4.8 | $5.25 | $26.25 |
Questions
The fine print, in plain words
- Is the Free plan really free?
- Yes for access: up to 3 members can build, run, and govern workflows at no cost, and every org gets 500 runtime-minutes of compute free each month. Beyond that, what you run draws from a prepaid credit balance — and model tokens are pay-as-you-go from the first token (no free token credit). The Free plan has no seat fee; Team adds seats at $20 each.
- What actually costs money?
- Usage, plus seats once you grow past the Free plan's 3 members ($20 per seat on Team). A run draws credits for the runtime it holds a machine plus managed model cost. Viewing, building, managing, roles, secrets, and audit are free. Bring your own model key and tokens are free from us too.
- What exactly is a runner-minute?
- A minute a run holds a machine while it's active. We meter the wall-clock time a run occupies a runner (including time it spends waiting on a sleep or a child run), and nothing when no run is executing. Larger machines are billed proportionally more, at $0.05 per vCPU-minute. Every org gets the first 500 runtime-minutes each month free.
- Do I pay while an agent is sleeping?
- If a run holds a machine across a sleep or a wait, that machine is reserved for it, so it's metered like any active minute. For long scheduled gaps, start a fresh run on a schedule instead of sleeping in-process — and your prepaid balance is a hard ceiling either way.
- Do you mark up model tokens?
- On the managed lane, we bill the model provider's upstream cost plus a 5% margin, metered as real cost, so a pricier model is just covered fairly rather than by a flat per-token rate. Bring your own key and we charge nothing on tokens at all; you settle with your provider directly.
- Are these prices final?
- We're pre-launch. The model (access plus prepaid credits for usage, with your-key-pays-nothing) is set; exact figures may move before general availability, and committed-use discounts are on the way.