Usage limits and Compute Units (CU)

Usage limits are one of the most important things to understand about Scenario: every generation, training run, and AI tool you use draws from a pool of Compute Units (CU), and CU is what your plan, your team's spending controls, and your API keys are all built around. This article explains what a CU is, where usage limits apply, how to configure them in the webapp, and what happens when a limit is reached.

What is a Compute Unit (CU)?

A Compute Unit is Scenario's metering unit for AI compute. Actions that consume CU include:

• Image generation (including with custom/trained models)

• Model training

• Background removal, upscale, pixelate, vectorization, embed, and segmentation

• Skybox and texture generation

• Detection and AI content-safety checks

• Assistant/agent messages

The CU cost of a generation depends on the model, image resolution, number of steps, whether you're using a custom model or LoRA, and batch size. Wherever you generate, Scenario shows the CU cost right on the button before you click, so there are no surprises, and it flags unusually expensive generations as "High cost."

What "current billing period" means

Usage limits are always measured "for the current billing period," and that period isn't the same for everyone:

• On a paid plan, your billing period runs on your own subscription's monthly cycle, anchored to the date you subscribed (or last upgraded), not the calendar month. If you subscribed on the 15th, your CU resets on the 15th of every month. You can see your exact reset date in Team Settings → Plans → Manage, labeled "Resets {date}" under your monthly allowance.

• Upgrading your plan resets your billing period immediately: your cycle restarts on the day you upgrade, and your reset date moves to that day going forward. Downgrading doesn't cut your current period short: the change takes effect at your existing cycle's next reset, so you keep what you already have until then.

• Purchased top-up CU is separate from the billing period entirely: it doesn't "reset," it expires on its own fixed schedule (shown as "Expires {date}" rather than "Resets {date}"), and is spent before/alongside your monthly allowance.

The four places a usage limit can apply

Your team has one pooled CU balance for the current billing period. On top of that pool, admins can optionally set caps that restrict how much of the pool a subset of the team can draw from. There are four levels:

1. Team plan allowance – the monthly CU allowance included in your subscription. This is the overall bucket everything else draws from.

2. Per-member limit – an optional cap an admin sets on an individual team member, so one person can't use more than their share of the team's CU.

3. Per-project limit – an optional cap on a specific project, so one project's usage can't consume the whole team's balance.

4. Per-API-key limit – an optional cap on an individual API key (covered in our API keys and scopes article.

Where to view and configure usage limits in the webapp

• Organization Settings → Plans → Manage: shows your current CU balance ("Compute units available"), broken into your monthly plan allowance (with its reset date) and any purchased top-up CU batches. From here you can:

  • Buy compute units: a one-time top-up purchase.

  • Auto-Refill: automatically buy a set amount of CU whenever your balance drops below a threshold you choose, so you don't get interrupted.

  • Default usage limit: set a CU cap that automatically applies to every member and API key that doesn't have its own individual limit.

• Organization Settings → Members: admins see each member's usage as a bar ("used / cap CU"), which turns amber near 80% and red at 100%. Click a member to set, raise, or remove their individual cap.

  

You can also set a common usage limits for all your members.

• Organization Settings → Projects: same usage bar per project, editable by team admins or that project's own admin.

  

• Organization Settings → API Keys: set an optional CU cap when creating a key, or edit it later from the key's row.

  

What happens when you hit a limit

Per-member or per-project cap reached: generation is blocked with a message explaining that the cap was reached, and that a team admin can raise or remove it from Members or Project settings.

Good to know

CU caps are independent of scopes. An API key's usage limit only restricts how much it can spend; it has no effect on which endpoints the key is allowed to call (that's what scopes are for, see API keys and scopes article.