Costs
The cost of each run, shown on the run
Model, tokens and estimated cost show on every run, live. You know what an agent consumes; totals per agent or per team are not there yet.
Estimates
A reliable order of magnitude, honest about its limits
The cost shown is an estimate computed from the model's prices, not an invoice. When a price is missing, the app says so instead of letting you think it was free.
- An asterisk flags a partial cost: at least one call used a model without pricing, so the real cost is higher.
- A model with "No pricing" counts as $0; its price is set through the API.
- With a personal key or subscription, the provider bills you, on its own terms.
Guardrails
Cost informs, limits protect
A run's estimated cost does not stop it halfway. What is capped is each run's duration, memory and CPU, and the pace of launches in the organization.
- Each run gets at most 30 minutes, 1.5 GB of memory and 2 vCPUs.
- An organization launches up to 200 runs per minute and runs 50 at the same time.
- "Cancel" stops a run heading the wrong way before the limit.
On the cloud
Model credits, at a glance
On Appstrate Cloud, the sidebar shows the model credits used, for example the $5 offered on Free. When paid plans open, the "Billing" page will gather the plan and the usage. When you self-host, none of this: you pay your model provider directly.
- The "Credits" gauge in the sidebar.
- The "Billing" page: plan, usage and storage, when paid plans open.
Today, and what comes next
Cost tracking happens run by run. Totals and budgets are on our roadmap.
Available today
- The estimated cost and tokens of each run, live while it runs.
- The per-turn breakdown: context, output tokens and latency.
- The context gauge, to see an agent getting close to its limit.
- On the cloud, the credits gauge and the "Billing" page.
On our roadmap
- A total per agent, per period, per space or per person.
- A cost column in the runs list.
- A cost export.
- A usage history screen.
Meanwhile, every run keeps its model, its tokens and its estimated cost: a manual check is still possible.
Budgets for models, per team and per agent
Set a spending cap in dollars for the whole organization, a space, a person or an agent, separately for the chat and for agents. A budget that is reached blocks what it covers, and the refusal names it.
- An alert to the owner at 80%, with an adjustable threshold.
- Per day, week or month, with exceptions per person.
- A Claude or ChatGPT subscription counts in dollar equivalent.
- In the open-source core: no paid module needed.
Frequently asked questions
Is the displayed cost an invoice?
Can I see total cost per agent or per team?
Who pays when I use my own key or subscription?
How do I avoid a bad surprise?
Do credits exist when self-hosting?
Let's see what it would look like on your side.
Try it on your own in a few minutes, or take thirty minutes with us to scope a first use case.
Let's talk about your case
Thirty minutes with us to scope a first use case and its gain.
Read the code
Apache 2.0. Your technical team can read it all, install it all, keep it all.