Skip to navigation
UploadingAI Image Editor

AI Operations billing

Beta

AI Image Editor is in public beta, so costs and limits may change.

AI Image Editor usage is billed in AI Operations, separately from regular Operations. Pro includes 5,000 AI Operations per month and Business includes 25,000, then overage costs $0.80 per 1,000. The number of AI requests per 24 hours is capped separately. See Limits and moderation. You can follow usage in Project > Analytics.

Cost per action

Action1 MP2 MP4 MP
Generate200280440
Edit280440760

Generate and edit cost 200 + 80 × (total MP − 1) AI Operations, where MP is megapixels. Total MP is the output size plus the size of each input image, rounded up to whole MP per image. A generation has no input. Input images larger than the output are scaled down to the output size, so a 2 MP edit of a larger photo counts 4 MP. Smaller input images keep their size. The Edit row assumes one source image at least as large as the output.

The output size is a project setting: 1, 2, or 4 MP, with 1 MP by default. Change it in Project settings > AI > Output image size.

Upscale costs a flat 50 AI Operations per request, at any size. It’s available through the API only.

Plans

AI Image Editor is available on Pro, Business, and custom plans, not on Trial or Free.

PlanIncluded per monthAt 1 MP, that’s aboutOverage
Pro5,000 AI Operations25 generations or 17 edits$0.80 per 1,000
Business25,000 AI Operations125 generations or 89 edits$0.80 per 1,000
Custom25,000 AI Operations by default125 generations or 89 edits by default$0.80 per 1,000 by default

On custom plans, the contract can change these defaults.

Each month, included AI Operations are used first, and overage applies only after they run out. Overage is billed in whole thousands, rounded up, like regular Operations. Unused AI Operations don’t carry over to the next month.

What isn’t billed

You’re billed only for images that are actually generated, edited, or upscaled.

  • Requests declined before processing cost nothing: a limit is reached, AI is turned off, the input is invalid, or the token is missing or invalid.
  • Failed generations cost nothing: a blocked prompt, a timeout, or the service being unavailable.

Requests that aren’t billed don’t count against your workspace request limit or your token’s operations.

Tracking usage

  • Project > Analytics: the AI Operations tile, and the AI Operations card split into Generation, Editing, and Upscaling. Both follow the Analytics date range. AI Operations appear there with a delay.
  • Billing: the plan card shows included AI Operations per month, and the invoice lists AI Operations as a separate line.