Usage

Every Roflow plan comes with a monthly AI allowance — this page explains how that allowance is spent, where to check what's left, and how to make it last.

How AI usage is measured

Your AI usage is based on the amount of work the AI does for you, not a flat charge per message. Under the hood the agent reads parts of your project, thinks through the task, and writes code or answers — and the more it has to read and write, the more of your allowance a request uses.

In practice that means:

  • Bigger, more complex tasks use more than simple ones. "Rename this variable" costs far less than "build a full save system and wire it into my shop."
  • Long conversations cost more per message. As a task grows, the agent carries more context along with it, so later messages in a very long chat draw down more than the first few. (You can see this in the task header, which shows how much of the context window a task is using.)
  • The model you pick matters. More capable models do more work per request and use more of your allowance.

Model choice and usage

Inside the agent's composer, the model picker lets you choose a tier for the current task, and each one is labeled with roughly how much of your allowance it uses:

ModelBest forUsage
FastQuick edits and simple tasksLow
BalancedStrong all-round coding (and it can see images) — the defaultBalanced
MaxComplex reasoning and debuggingHigh

The picker notes that higher-tier models use more of your monthly allowance, so pick the lightest model that can handle the job. See Talking to Roflow for more on choosing a model.

Where to see your usage

Roflow surfaces your usage in a few places, so you can keep an eye on it wherever you're working.

In the agent chat

When it's enabled, a usage bar just above the message box in the agent panel shows your plan (Free, Pro, or Pro+) and the percentage of your monthly allowance you've used for the current billing cycle. The bar changes color as you get closer to your limit so you notice before you run out.

Each task also has an expandable task header that shows the running token counts, the cost so far, and how much of the context window that task is using — a useful signal that a conversation is getting long and would be cheaper to continue as a fresh task.

In Settings

Open Settings from the gear (Manage) or account icon in the left navigation. Under Plan & usage you'll see:

  • Your current subscription plan.
  • Usage this period as a percentage, with a meter that highlights when you're near your limit.
  • How many requests you've made this period.
  • When your allowance resets.
  • Upgrade (or Manage plan) and View usage links that open roflow.co.

If usage details can't be loaded, Settings tells you so and points you to your web dashboard.

On the web

The View usage link opens your account on roflow.co, which shows a Monthly AI Usage card for the current billing cycle: the percentage used, how much is remaining, the reset date, and a 6-month trend chart so you can see how your usage changes over time.

When you're running low

As you approach your limit, the usage bar and the Settings meter change color to warn you. Nothing stops yet — this is just a heads-up so you can slow down, switch to a lighter model, or upgrade before AI features pause.

When you run out

If you use up your monthly allowance, AI features pause until it resets — but the rest of the IDE keeps working. You can still edit code by hand, sync to Studio, run playtests, and run commands; only the agent's AI requests are on hold.

When you hit the limit, the chat shows a banner telling you your limit was reached and how much you've used, along with two buttons:

  • Upgrade Plan (or Buy Credits) — opens roflow.co so you can move to a higher plan and get more allowance immediately.
  • Retry Request — resend the request once your allowance has refreshed (or after upgrading).

To get going again right away, upgrade your plan; otherwise your allowance refreshes on the reset date shown in Settings. For more on AI-related errors and limits, see AI Issues.

Tips to make your usage go further

A few habits stretch your allowance a long way:

  • Match the model to the task. Use Fast or Balanced for edits and everyday coding, and save Max for genuinely hard debugging or reasoning. Running everything on the most capable model burns through your allowance fastest.
  • Plan before you build. Using Plan mode or /deep-planning to agree on an approach first means fewer wasted, rebuilt-from-scratch attempts. See Planning Your Game.
  • Give good context up front. Mentioning the right files, folders, or problems with @ and attaching images means less back-and-forth, and less back-and-forth means less usage. See Getting Great Results.
  • Work in focused stages. Smaller, well-scoped tasks are cheaper and easier to review than one sprawling request. See Building in Stages.
  • Start a new task when you switch topics. Because a conversation carries its history along, a long chat costs more per message. Beginning a fresh task (or using the /smol command to condense the current one) keeps the context — and the cost — lean.
  • Edit and resend instead of starting over. If a request didn't land right, edit your earlier message and resubmit rather than re-explaining everything from scratch.
  • Polish the prompt first. The Improve prompt button rewrites your draft into a clearer request, which helps the agent get it right on the first try instead of after several corrections.

Where to go next