Billing

Forginate bills organizations on a simple model: a base plan plus metered model usage — or, if you bring your own API keys, just the base plan with token costs passed straight through to your provider. This page explains plans, metering, budget caps, and where to see every dollar.

The two cost tracks

Platform tokens (default)

Forginate supplies the model credentials. Every model call your projects make is metered:

• Usage is priced from the actual tokens consumed per model, at published per-million-token rates, with your plan's markup applied.
• Paid plans include a monthly token allowance; usage beyond it is billed as overage on your invoice.
• You never see or manage API keys.

Bring your own keys (BYO)

Add your organization's own Anthropic, OpenAI, or OpenAI-compatible API keys under Settings → Providers:

• Token costs go directly to your provider account — cost passthrough, no markup from Forginate.
• You still pay the base plan.
• Keys are write-only: the plaintext is accepted when you save it, stored encrypted, and never displayed again — reads show only a last-4-characters hint. Use Test connection to verify a key works, and mark one provider as the org default (the vibe editor uses your default Anthropic provider).

You can mix tracks — e.g. BYO for heavy build work, platform models where your keys don't cover a model.

Plans

Your organization is on a plan that sets the monthly base price, the included token allowance, the token markup, and limits such as how many projects you can run. Manage it from the Billing page:

• Upgrade / subscribe via a secure Stripe checkout (org owner only).
• Manage the subscription — payment methods, plan changes, cancellation — through the Stripe customer portal, linked from the Billing page.
• Invoices — your recent invoices are listed on the Billing page.

Where to see your spend

Cost visibility is a first-class feature, not a surprise at month-end:

• Per run. Every agent run shows its token usage and dollar cost, updated live while the run executes — not just after it finishes.
• Per stage. A request's detail page breaks total spend down by pipeline stage (planning, build, audit rounds), so you can see where the money went.
• Per estimate. The TOC screen shows a cost estimate before you approve document generation, priced from live model rates.
• Org-wide. The Usage page rolls up your organization's usage, and the Billing page shows current-month totals.

Budget caps

Runs don't get a blank check:

• Every run has a per-run cost cap from your plan. A run that reaches its cap stops instead of continuing to spend.
• The integration phase's audit-and-fix loop works inside a budget envelope. If the envelope is exhausted before the audit is green, the work escalates to you rather than looping silently.
• You can raise the budget mid-run — the "up the cap" control adjusts the live envelope without resetting progress, letting the work continue under a larger limit you've explicitly chosen.

Delayed runs don't cost extra

When a model provider is overloaded, runs wait and retry in place rather than failing and restarting. Waiting costs nothing; the run resumes its exact position when the provider recovers, with zero re-spend.

Model tiers

Different pipeline stages use different model tiers — heavier models where quality matters most, lighter ones where it doesn't — and each tier's per-token rate feeds both the estimates you see up front and the metering that lands on your invoice. The same rates are used consistently everywhere: estimate, live run cost, and bill.

Billing FAQ

What happens if a run fails? You pay for the tokens actually consumed; failed runs don't bill for work that never happened.

Can a viewer see costs? Cost readouts on runs and requests follow project access; billing management (checkout, plan changes) is owner-only, while the Billing page itself is visible to org members.

Do BYO orgs see cost numbers too? Yes — usage is still tracked and displayed so you know what your own keys are spending; it just isn't billed by Forginate.