Pay per actor.Not per feature.
One subscription per organization covers the whole platform. The floor follows who operates it, people and agents with identities of their own, and usage above what the platform fee includes is metered on runs and requests. One published plan, the same rates for every customer.
US$ 49per month
The whole catalog, console, API and MCP for the organization. Includes 100,000 workflow executions, 1 million requests, 10 GB of storage and 5 end users every month.
US$ 20per person / month
A named person on your team. Every action is recorded in their name.
US$ 5per agent / month
An AI agent or integration key with an identity of its own. Tied to a named person, it costs half.
Começar, Equipe and Escala on /start are prepaid bundles of this model — platform, seats, retention and Lumen — not a second price list.
Then you pay for what you ran.
Above the volume the platform fee includes, two meters: the runs the engine executed, and the calls that reached the platform.
US$ 150per million
One engine run of a workflow, governed, attributed and retained. 100,000 included each month. An agent call consumes this meter: if it fans out into several runs, each run counts.
US$ 5per million
Inbound platform traffic: API calls, webhook and ingest hits. 1 million included each month.
What you keep, and for how long.
Run history has a window you choose. Lumen, the observability product, is priced on its own.
US$ 0.60per GB / month
Run history beyond the 10 GB the platform fee includes.
US$ 8 to US$ 140per month
How long run history is kept: 30 days, 90 days, 365 days, 5 years or 7 years. Every tier includes 10 GB.
US$ 50per month
Errors, traces, logs and metrics for your operation.
US$ 250per month
Higher volume and longer retention.
US$ 0.40per GB / month
Telemetry kept beyond the tier.
US$ 4per million
One entry is one log, trace or metric record.
Three ways to take a seat.
An actor is any identity that acts in your operation — human or not. It is the only unit of billing.
User seat
One person on your teamAccess to the console, the API and MCP with an identity of their own. Every action that person triggers is recorded in their name.
Key seat
An agent or autonomous integrationA key with an identity of its own for AI agents, integrations and automations. No shared identity: every action is attributable to whoever ran it.
Delegated agent
A key tied to a personWhen an agent acts under the authority of a named person on your team, the key seat costs half.
Everything included, for every customer.
No capability sits behind a higher plan. What the platform does, it does for everyone.
- The entire capability catalog — nothing gated behind a tier
- Every product: deploys, runners, Kubernetes, networks, registries, webhooks and more
- Governance and auditing on every run: each transition recorded with actor, state and timestamp
- All three surfaces: console, API and MCP
- Execution on your own cloud — infrastructure cost is yours, billed directly by your provider
- The same plan for every customer, from a team of two to a team of two hundred
Predictable by construction.
A fixed floor, then usage
Seats and add-ons are prepaid. Executions and requests are metered above the volume the platform fee includes, so a quiet month costs the floor and a busy one pays for what it used.
Upgrades apply now, downgrades at renewal
Adding seats or capacity takes effect immediately, prorated. Reducing schedules the change for your renewal date, and can be cancelled until then.
History with configurable retention
Pick the window that suits you, from 30 days to 7 years. Each tier includes 10 GB; storage past that is metered per GB.
Observability is an add-on
Lumen is priced separately: a monthly tier, plus telemetry storage and entries. Integration traffic bills as requests, never twice as entries.
Access tied to the contract
A key seat lasts for the contracted period. On expiry, the key drops to read-only until renewal.
Sign in, confirm this plan, then connect MCP.
The console shows the same amounts. You subscribe there; after payment the home page is the MCP wizard — URL, clients, and a first whoami prompt. If you have a coupon, add it on Stripe before you pay.
Sign in to subscribeWhat people ask most.
How much does Orkestia cost?
US$ 49 per month for the organization, plus US$ 20 per person and US$ 5 per agent or integration key. That includes 100,000 workflow executions, 1 million requests, 10 GB of storage and 5 end users each month. Above that, executions are US$ 150 per million and requests US$ 5 per million. A team of three with one agent pays US$ 114 per month before usage.
Is the price negotiated per customer?
No. There is one published plan and every customer pays the same rates. No tiers, no bespoke deals, no capability behind a higher plan.
What counts as an execution, and what counts as a request?
An execution is one engine run of a workflow. A request is one inbound call to the platform: an API call, a webhook or an ingest hit. An agent call consumes the execution meter, and if it fans out into several runs, each run counts. The US$ 5 agent seat is the identity charge only.
Is there a free tier?
Not yet. Sign in, subscribe on the published plan, then connect MCP. If you have a coupon, add it on the Stripe page before you pay.
Do I need an enterprise plan?
No capability sits behind a higher tier. What varies is how many people and agents operate, how much you run, and optional retention and Lumen.
Where do workflows run? Who pays for the infrastructure?
On your own cloud. Orkestia is the control plane; the compute, network and storage for your resources stay in your account, paid by you directly to your provider.
How are AI agents billed?
Each agent takes a key seat of its own at US$ 5 per month, or half that when the key is tied to a named person on your team. What the agent runs bills on the execution meter, like any other run.
Does observability cost extra?
Yes. Lumen is an add-on: US$ 50 per month for Pro or US$ 250 for Enterprise, plus US$ 0.40 per GB of telemetry storage and US$ 4 per million entries, where an entry is one log, trace or metric record.
What about the end users of the app I build on Orkestia?
End users of your apps have their own seats, separate from your team's seats. The platform fee includes 5, and more are bought in packs as your base grows.
Can I predict the invoice?
The floor is fixed and known: platform plus seats plus any add-ons. On top of that you pay for executions and requests above the included volume, so the variable part follows how much you actually ran.