← All help topics

Billing and token usage

The “Billing” section (above “Profile”) shows actual token usage across all your projects and helps plan the budget. The project Overview shows a weekly estimate and the actual figure.

1Where the numbers come from

After each model response the service stores the actual token usage returned by the provider:

  • 📊 billing shows real data, not a guess;
  • 🧩 input + output are counted;
  • 🖼️ images are billed separately (per image) and aren't counted in tokens.

2On the Overview: weekly actual and forecast

The “Tokens per week” card shows two different numbers:

  • 📊 in large type — the ACTUAL usage over the last 7 days;
  • 🗓️ below it — “≈ forecast by the publishing plan”.

The forecast = average tokens per post (from project history) × expected posts per week (from active plan days and templates), capped by the weekly limit — roughly what regular feed posting in “Plan and post queue” would consume.

The forecast does NOT include one-off and manual actions, so the actual can easily be HIGHER than the forecast. The forecast excludes:

  • ✨ the “Generate drafts” button and text refinement;
  • 🖼️ image regeneration;
  • 🔍 trend computation in “Trends”;
  • 🧠 “Improve project”;
  • 🔗 importing from a site/channel.

For example, if in a single day you manually generate and regenerate posts and images many times, the daily actual can exceed the weekly forecast. The forecast is a “steady-state” guideline, not a cap or a bill.

3Period report and daily journal

In the “Billing” section pick a date range (or a preset “7 days” / “30 days”) and click “Show”. The report includes:

  • 📊 total tokens and number of requests;
  • 🤖 a breakdown by provider;
  • 🧩 a breakdown by project (with providers inside each one);
  • 🗓️ a daily journal detailed by action (posts from sources, trends, templates, improvement, etc.).

This shows which project and which provider spends how much for any period. Under each provider and each project there's a mini per-day chart: colored bar segments show what the tokens were spent on each day (by purpose for a provider, by providers for a project), with an exact breakdown on hover.

4Estimated cost and where the price comes from

Next to tokens, an ESTIMATED cost is shown using approximate public provider prices. Key things to understand:

  • 💳 the cost is computed per model (Claude Opus and Haiku differ 5×), including prompt caching — the part of a request read from the cache costs ~10% of the input price; the “By provider” breakdown lists the models used (older records without a model use the provider rate);
  • ⚙️ the prices in the service are a guideline and may go stale (there's no universal way to auto-fetch current prices across all providers);
  • 🔍 it's a very rough guideline: the exact amount depends on your plan, region and specific model — check it in the provider's console;
  • ⛔ for a custom model the estimate isn't computed.

5🎬 Video cost (image animation)

Video is billed NOT in tokens but per clip (by seconds), so it's counted separately:

  • 💳 in «Billing» the video providers (Veo / Runway / Kling) appear as a separate line — the number of clips and an approximate cost — and are included in the total «Estimated cost»;
  • 📊 the project «Overview» has a «🎬 videos this week ≈ $…» line;
  • 🔍 it's an approximate estimate (depends on duration, resolution and your provider plan).

To avoid waste, cap auto-animation with the «videos per day» limit in «Project info» — see «Animating images into video».

6Token budget and alert

In project settings you can set a “Token budget per 7 days”:

  • ⏱️ when the actual weekly usage exceeds the budget, you get a notification in “Notifications” — it isn't duplicated but updated;
  • 🟢 a value of 0 means no budget.

This helps you stay within limits.

7How to reduce usage

A few ways to reduce token usage:

  • 📝 keep the context concise — shorter extra/marketing context and files, or enable “Economy mode”;
  • 💳 assign cheap models to frequent utility tasks in “AI providers and API keys”;
  • ⏱️ set generation limits;
  • 🔒 pause inactive projects.

See “How generation works” for details.

Didn't find an answer? Write to us — we'll help.

Contact support