Plans & limits
Boardwalk is free to build on and run locally, forever, on the open-source engine. A subscription turns on the hosted side: durable runs on managed runners, triggers that fire in the cloud, managed inference, and self-hosted runners attached to the hosted control plane. This page is the reference for what each plan includes and how the limits behave; prices and per-unit rates live on the pricing page.
The plans
| Free | Solo | Pro | Team | Enterprise | |
|---|---|---|---|---|---|
| Price | $0 | $20/mo | $100/mo | $40/member/mo | Custom |
| Members | Up to 3 | 1 (just you) | 1 (just you) | Unlimited, every member billed | Volume-tiered |
| Hosted runs | No (local runs only) | Yes | Yes | Yes | Yes |
| Agent-hours / month | None | 25 | 150 | 25 per member, pooled | Committed |
| Managed tokens / month | None (BYO key only) | $20 at cost | $100 at cost | $40 per member, pooled | Committed packs |
| Web searches / month | None | 500 | 2,000 | 1,000 per member, pooled | Metered |
| Concurrent hosted runs | None | 3 | 10 | 3 per member, pooled | Dedicated capacity |
| Storage | 5 GB (hard cap) | 20 GB | 100 GB | 100 GB + 25 GB per member | Custom |
| Self-hosted runners | None | 1 | 3 | 10 per org | Negotiated |
| Triggers | Deployable, paused | Active | Active | Active | Active |
| Governance | Roles + audit log | Roles + audit log | Roles + audit log | + require-2FA, email domain policy | + SSO/SAML, SCIM, custom retention |
Solo and Pro are single-member plans. The moment an org adds anyone else it moves to Team, where every member is billed regardless of role and allowances pool org-wide. Enterprise adds committed usage packs, a self-managed deployment option, dedicated capacity, and named support with an SLA; talk to us.
Allowances & the monthly reset
Every paid plan includes monthly allowances: agent-hours, a managed-token pool, web searches, and storage. Two semantics to know:
- They reset monthly, on your billing anniversary. Each allowance refills in full at the start of each billing cycle and unused amounts never roll over. Prepaid credits are separate and never expire.
- You always see them in real units. The dashboard and
boardwalk usageshow gauges like "14.2 of 25 agent-hours" and "$3.10 of $20 managed tokens", plus the date the allowances reset. There are no rolling windows and no abstract percentages.
An agent-hour is one hour of active run time on a standard 1-vCPU runner. Larger machines draw the allowance proportionally: an hour on a 4-vCPU machine uses 4 agent-hours, matching the per-vCPU compute meter. Suspended time (a long sleep, a human-input gate, waiting on a child run) releases the machine and draws nothing.
The managed-token pool draws at the model's at-cost rate and works on any surface, including local boardwalk dev. Bring-your-own inference keys bypass the pool entirely and cost $0 from Boardwalk on every plan.
Concurrency & queueing
Each plan caps how many hosted runs execute at once: 3 on Solo, 10 on Pro, and 3 per member pooled org-wide on Team. The cap never fails a run:
- A run that starts while the org is at its cap queues and starts automatically when a slot frees, in FIFO order.
- A resumed run (waking from a sleep or an answered human-input gate) takes priority over queued new starts, so a workflow that suspended never gets stuck behind its own backlog.
- Queued runs are visible in the dashboard with their queue position; nothing errors and nothing is dropped.
Trigger pausing & resume
Cron schedules, webhooks, and workflow-to-workflow triggers deploy on every plan, including Free. Whether they fire depends on the plan:
- On Free, triggers deploy in a paused state. The workflow, its schedule, and its webhook URL all exist; subscribing activates them.
- On downgrade or when a subscription lapses, triggers pause the same way. The pause is loud: a dashboard banner, a CLI notice, and an email. Nothing is deleted, and run history stays.
- On resubscribe, resuming paused triggers is one click.
Spend caps
You can set an optional org spend cap: a ceiling on overage spend per billing cycle. A run that would start past the cap does not start until you raise the cap or the cycle resets, and the block links straight to the setting. It is the guarantee that a runaway loop can cost at most what you signed up for.
Overage, credits & the fee
When an allowance runs out, the same usage keeps working as overage, drawn from prepaid credits at the public rates, identical on every plan: compute at $0.05 per vCPU-minute, managed tokens at the model's source cost, storage at $0.25 per GB-month, and web search at $0.008 per search (full tables on the pricing page).
- Usage carries no markup. The only fee is 10% on a prepaid credit top-up: a $50 top-up costs $55 and adds $50 of credit. It covers card processing and the upstream credit fee on routed tokens, which is what keeps per-model rates at cost.
- Purchased credits never expire, and when they run out, runs stop rather than producing a surprise bill.
- Free has no metered usage at all: its 5 GB storage is a hard cap and it never bills.
Self-hosted runner limits
Paid plans include self-hosted runners: your machines attached to the hosted control plane. Solo includes 1 runner, Pro 3, and Team 10 per org; Enterprise caps are negotiated. Registering a runner past the plan's cap is rejected at registration.
Billing follows where the work happens: compute on your own runners is never billed and draws no agent-hours, while brokered usage the run makes through the control plane (managed tokens, artifact storage, web search) bills normally against allowances and credits.
For running the whole platform on your own infrastructure with no account and no plan, see Self-hosting.