Pricing
Bring the AI agents and subscriptions you already pay for. Ozthra AI Relay charges for the local control layer—orchestration, guardrails, approvals, and audit—and never meters your model usage. Start free; scale into team governance when you are ready.
Ozthra AI Relay is live in early access — signup is open, and every new workspace is reviewed and approved, usually within a day. Start on the Free plan, or go straight to a paid plan with self-serve billing. Early-access teams lock a founding rate.
Plans
Every plan is bring-your-own-agent and bring-your-own-subscription. You only pay Ozthra for orchestration, governance, and team adoption.
Free
$0 forever
Small teams — get started on your machine
Pro
$12 / user · mo · 5-seat min
Power users & small teams
Team
$29 / user · mo · 5-seat min
Teams that need governance
Enterprise
Custom from $20k / yr
Security-reviewed organizations
Prices in USD, billed monthly via Stripe (5-seat minimum on paid plans). Bring your own agent CLIs, subscriptions, and API keys on every plan—Ozthra AI Relay does not resell tokens or meter your model usage.
Pricing principles
The thing other agents are starting to meter, we refuse to. Because local orchestration costs us almost nothing to run, we charge for the control layer and leave your model usage entirely to you.
We charge for: orchestration, guardrails, approvals, audit, templates, team governance, and implementation.
We never charge for: your tokens, your model usage, or your local runs—and we never mark up your API keys.
You bring official subscriptions and keys, used under your own account and control.
Cloud runner
Everything runs on your machine today, with no per-run or per-token billing. Team is fully uncapped; Pro includes a 2,000 runs per user / month fair-use allowance and Free 20 runs per user / day—anti-abuse guards, never billing meters. When the optional cloud runner ships, it will be a separate, opt-in add-on priced on the compute it uses, with an allotment included on paid plans—so the line between "your machine" and "our compute" stays clear and honest.
Opt-in only—off by default.
Usage-based, billed on compute actually used.
Same guardrails, approvals, and audit as local runs.
Implementation
A fixed-scope, two-week engagement—$7,500—to review your workflows, set up a governed pilot, and produce an adoption plan. Sold separately from product subscriptions.
FAQ
No. Ozthra AI Relay never resells tokens and never marks up model usage. You run official CLIs and APIs under your own account, subscription, or API key. We charge for the control layer—orchestration, guardrails, approvals, and audit.
No per-run or per-token billing—your bill never grows with the number of workflow runs, and tokens are never metered on any plan. Team has no run caps at all. Pro includes a generous fair-use allowance of 2,000 runs per user per month, and Free allows 20 runs per user per day—both are anti-abuse guards, not billing meters; nothing is ever charged per run. The only usage-based pricing will be the optional cloud runner, which you turn on deliberately.
Paid plans are billed monthly per seat through Stripe, with a 5-seat minimum on Pro and Team. Payment happens in Stripe's hosted checkout and billing portal—Ozthra never stores your card details. Your invoices are available in the app with PDF downloads.
Upgrades and added seats apply immediately with a fair prorated charge for the rest of the billing period. Downgrades and seat reductions take effect at the end of the current period—you keep what you paid for until then, and you can undo a scheduled change any time before it lands.
Yes. Cancelling takes effect at the end of the current billing period: you keep full access until then, and your workspace then moves to the Free plan with your data intact. You can undo the cancellation any time before the period ends.
From Billing in your workspace: open the Stripe billing portal to update your card or billing details, and view or download any past invoice as a PDF. Plan and seat changes are self-serve from the same page.
Local orchestration runs on your machine and costs us almost nothing to operate, so we pass that on. We make money from team governance and implementation, not from metering your runs.
The optional cloud runner is a separate, opt-in add-on, priced on the compute it actually uses, with a monthly allotment included on paid plans. It is only billed when you choose to run a workflow on Ozthra's cloud instead of your own machine. Final rates are set at launch.
Per seat for the control layer—not per run. Pro and Team are flat per-user prices billed monthly; Enterprise is custom. If you prefer a flat team price instead of per-seat, ask us during a sales conversation.
Yes. The AI Adoption Sprint is a fixed-scope, two-week engagement to review your workflows, set up a governed pilot, and produce an adoption plan. It is sold separately from product subscriptions—see Services.
Early access
Ozthra AI Relay is live in early access on macOS, with Windows in development. Bring your own agents and subscriptions, keep guardrails and audit on your machine, and never pay a second usage meter.