DISPATCH · NO SUBSCRIPTION
pay for the work you run.
nothing you don’t.
Inference is pay-as-you-go: no trial clock and no included allowance. Top up prepaid credit, route your agents through LumenFlow, and pay a flat 10% governance fee on the work you actually run. Team and Enterprise plans layer on top for more workspaces, longer retention, and enterprise controls.
How you pay
Three moving parts, and that’s the whole model.
Prepaid inference credits
Top up a credit balance whenever you like. Governed model calls draw down from it as you run them — no plan to pick, no monthly reset, no commitment. Your balance is yours until you spend it.
10% governance fee
One headline fee on the work you route through LumenFlow. It pays for the routing, policy checks, approvals, audit trail, and evidence receipts wrapped around every call — the governance, not just the tokens.
Compute is its own credit
Managed sandboxes, browsers, and GPU runs draw from a separate compute credit balance. Inference and compute never share a wallet, so a heavy run never quietly drains the other.
Pay for what LumenFlow governs
Models, your own hardware, LumenFlow-run compute, or enterprise evidence — the fee is for the governance, not for owning the infrastructure underneath.
| Source | You pay LumenFlow | How it works |
|---|---|---|
| Managed models | Token cost, through us, + 10% governance fee | We hold the provider keys and bill you the token cost plus the governance fee, drawn from your inference credit. |
| Your API key | Direct to provider + 10% governance fee | Bring your own model keys and route them through the LumenFlow gateway. Your provider bills you directly for tokens; LumenFlow charges the governance fee for governing the call. |
| Your compute | Yours — enterprise / BYOC contract pricing | Run governed work on machines you control. There's no published BYOC price list; connected-compute pricing for your hardware is set per enterprise contract. |
| LumenFlow compute | Compute credits — rolling out | Managed sandboxes, browsers, and execution runs draw from a separate compute credit balance as metering and checkout roll out — a governed execution substrate, not a marketplace. |
| Enterprise evidence | Contract pricing | The audit-trail export, legal hold, and the other compliance add-ons price into your enterprise contract rather than the per-token fee. |
Plans for scale
Inference stays pay-as-you-go on every plan. Plans add workspaces, retention, and enterprise controls on top.
Free
$0
- 1 workspace
- 7-day retention
One workspace with full governance features, built to earn your trust before you pay for anything.
Start freeTeam
$49/month
- 5 workspaces
- 90-day retention
More workspaces and longer retention for a team that has outgrown one owner.
Start freeEnterprise
Custom
- Unlimited workspaces
- 365-day retention
Unlimited workspaces, the longest retention we offer, and the enterprise controls below.
Talk to usIncluded monthly usage
Inference has no included allowance and no monthly reset. It is paid from prepaid credit. Free and Team workspaces include a monthly allowance of managed compute and of governed execution on your own hardware. It resets each month and, on Free, work stops when it is used.
| Plan | Managed compute | Browser | GPU | Governed execution |
|---|---|---|---|---|
| Free | 1 hour | 10 minutes | None | 1 hour |
| Team | 100 hours | 10 hours | 1 hour | 10 hours |
- Free: Work stops until the next month, or until you move to a larger plan.
- Team: Governed execution continues at $0.50 an hour. Managed compute is metered up to a monthly cap of $500.
These are current limits and may change with notice. They reset each month and do not carry over. Governed execution means work run on hardware you control; it is not AI usage.
Enterprise
Unlimited workspaces, a full year of evidence retention, and the controls a regulated buyer needs to say yes.
SSO toggle
Turn single sign-on on or off for the whole organization from one admin control.
Trust dashboard
A live surface showing evidence and control state for your organization — evidence, not a certification badge.
Evidence Vault
Every governed action lands a structured, timestamped record you can search and review. Receipts are hash-chained by default; turning on a dedicated Ed25519 signing key and an RFC-3161 timestamp witness makes new receipts independently signature-verifiable and their anchor externally witnessed.
Compliance export
A CSV or JSON audit-trail export of your governed activity and a control-to-framework attestation with honest per-control coverage. The export file itself carries no signature.
Legal hold
Flag evidence org-wide or per-record so it's excluded from the retention sweep for as long as you need it held.
365-day retention
The longest evidence window on offer, ahead of Free and Team.
Governed BYOK
Bring your own model keys and route them through the LumenFlow gateway — you keep the provider relationship, we keep the governance.
Pull-based connected compute
Your infrastructure pulls the work assigned to it. Nothing inbound is ever opened to reach it.
SSO authenticates through GitHub OAuth today. SAML and SCIM provisioning are shipped and enabled per organisation as soon as you connect an identity provider.
Ask about the compliance add-on: the audit-trail export (CSV or JSON) and legal hold are live today. The export file itself is not signed. A longer committed retention window and a response-time commitment for evidence retrieval are roadmap work we're building toward, not something we sell yet.
Compliance packs
Governance postures you can turn on per workspace — readiness controls for your own audit programme, not a certification badge.
GDPR — agent governance posture
EU GDPR posture for workspaces processing personal data of EU data subjects. Requires explicit approval routing and an explicit trust posture before agentic actions affect regulated data.
Custom
HIPAA — agent governance posture
US HIPAA Privacy + Security Rule posture for workspaces processing PHI. Requires explicit approval routing, explicit trust posture, and an authenticated connector path before agentic actions touch PHI.
Custom
Financial Services — agent governance posture
Regulated-financial-services posture (FINRA / SEC / PCI / MiFID II). Requires supervisory approval routing, explicit trust posture, authenticated connector path, AND an explicit model-source decision before agentic actions run against regulated workloads.
Custom
EU AI Act — agent governance posture
EU AI Act posture for workspaces operating AI systems in scope of the Act's high-risk obligations. Requires explicit human-oversight approval routing (Article 14), an explicit trust posture, and an explicit, approved model source before agentic actions run.
Custom
Sovereign residency — agent governance posture
Sovereign / data-residency posture for workspaces that must keep regulated data within a pinned jurisdiction. Requires an explicit, pinned residency policy (residency_pin) and an explicit trust posture before agentic actions run.
Custom
Pack pricing follows the Enterprise tier: contract pricing rather than a published self-serve list.
Data residency
Ask about region-specific hosting and model-routing commitments for your workspace. Automatic refusal of out-of-region calls remains under validation and is not a current guarantee.
Data residency region pin (Frankfurt)
Included on enterprise plans. Production hosting: EU (Frankfurt).
Custom
Residency pricing follows the Enterprise tier: contract pricing rather than a published self-serve list. Production hosting regions are listed from the platform hosting inventory.
Ready to govern agent work?
Create your account and start in a governed workspace. No card to enter until you choose to top up.