Skip to main content

Credits and Monthly Limits

Overview

Idealy tracks image credits with a single shared credit calculator across:

  • credit limit checks
  • header "credits remaining" meter
  • user/admin credit breakdowns

Monthly limits by tier

TierMonthly credit limit
Starter250
Plus500
Pro1000

Credit counting resets on the first day of each calendar month.

Image generation credit costs

CostModel/API type
0.5 creditsGPT1 Mini, GPT1 Mini Edit
2 creditsNano Banana 2, Nano Banana 2 Edit
4 creditsGPT1-HQ, GPT1-HQ Edit, GPT1.5-HQ, GPT1.5-HQ Edit, Nano Banana Pro, Nano Banana Pro Edit
1 creditAll other generation/edit model types (for example Flux, Flux-2, FluxKontextPro, Ideogram, Recraft, Imagen, Minimax, GPT1, GPT1.5, HiDream, Seedream, Seedream 4.5, Seedream 5 Lite, Krea, Qwen, Nano Banana, Bria, Nano Banana Edit, SeedDream 4 Edit, GPT1.5 Edit)

Operation credit costs (non-generation)

OperationCredit cost
Upscale image1 credit each
Vectorize image1 credit each
Background-removed vectorization1 credit each (counted as vectorization)

Color correction credit behavior

Color correction creates a new generated image and uses the source image's model type for credit accounting.

That means color correction cost is model-dependent:

  • 0.5 credit if source model is GPT1 Mini / GPT1 Mini Edit
  • 2 credits if source model is Nano Banana 2 / Nano Banana 2 Edit
  • 4 credits if source model is GPT1-HQ / GPT1-HQ Edit / GPT1.5-HQ / GPT1.5-HQ Edit / Nano Banana Pro / Nano Banana Pro Edit
  • 1 credit for other source models

What currently does not add a separate operation credit

  • Background removal/transformed-image upload by itself does not have its own dedicated credit counter entry.
  • A separate credit is only applied when you run a credit-counted operation after that (for example vectorization).

Limit enforcement

All generation and transformation endpoints use tier-based monthly limits (Starter/Plus/Pro), not a single global limit.

Practical usage tips

  1. Use 1-credit models for exploration passes.
  2. Use 2-credit models when you want a quality/cost middle ground.
  3. Reserve 4-credit models for finalists.
  4. Batch your upscales/vectorizations only on shortlisted images.
  5. Check your remaining credits regularly in the app header.