AI Credit Estimator
This feature is only available in paid plans. Learn More
Plan a monthly AI budget before you build. Enter how many end users you expect and how often each one uses each AI feature in a month. The result is an estimate, not a quote or a limit: your actual usage depends on your prompts, designs, and configuration.
Estimated monthly cost
$141–$829
141,200–828,600 AI Credits at list price
- Per end user
- $1.41–$8.29 per month
- AI Credits packs
- 3–17 packs ($150–$850/month)
By feature
- Whole-design requests
- $17.40–$57.60
- Text and block edits
- $105–$752
- Images generated
- $18.80–$19.00
| Feature | Typical | High (p90) |
|---|---|---|
| Whole-design requests | 174 | 576 |
| Text and block edits | 105 | 752 |
| Images generated † | 94 | 95 |
| Images edited † | 94 | 96 |
| HTML imports * | 639 | 672 |
| Screenshot imports * | 233 | 462 |
* Limited usage: these figures come from 20 recent operations, priced at current Claude rates. Validate them with your own requests.
† Measured on the earlier default image models. The latest editor revision defaults to GPT Image 2.5 Flare for new images and Nano Banana 2.1 for ordinary edits, so its image usage may differ. The public API and older editor revisions keep Gemini 3.1 Flash Image.
Typical is based on the median and high on the 90th percentile of credits per operation in production between September 18, 2026 and October 17, 2026 (HTML imports and Screenshot imports: September 29, 2026 to October 17, 2026), priced at the rates in effect on October 20, 2026. Both are scaled to spread each feature's billed routing, retries and failed requests across its operations. The high end adds up every feature's high figure, which a real month rarely reaches all at once.
How the estimate works
Each feature has two figures measured from recent production usage of paid Unlayer workspaces:
- Typical is based on the median operation: half of the operations used fewer credits.
- High is based on the 90th percentile: nine in ten operations used no more.
Routing calls, retries and failed requests also use credits, but they aren't operations of their own. Both figures are scaled by each feature's ratio of all billed credits to the credits of its completed operations, so every operation carries an average share of that extra work. Both figures count only requests that ran a model.
The low end of the range multiplies each feature's monthly operations by its typical figure. The high end uses the high figure.
Imports run much less often than other features. Their figures use the token sizes of recent imports, from up to the last 90 days and only since the import pipeline last changed in a way that affects cost. They're priced at the current rates of the model that runs imports. When only a few workspaces back them, the estimator marks them as based on limited usage. The high end adds up every feature's high figure, so treat it as a conservative budget rather than a likely month.
An AI Credit is a cost-based unit: at list price, 1,000 credits are worth $1. AI Credits are added in recurring packs of 50,000 credits for $50 a month, and unused credits don't roll over. The estimate shows the packs you would need at list price. Enterprise plans have negotiated AI allowances, so contact sales for Enterprise pricing. One-time welcome credits are not included. See Pricing for allowances, resets, and what happens when credits run out.
What changes your actual usage
The figures describe typical production requests, not yours. Expect higher usage than the typical figure when:
- Designs are large or carry embedded images. The Assistant processes the surrounding design, not just the selected block. Images embedded as data URLs can add substantial input and output.
- Requests carry more context. Brand Context, long conversations, and follow-up turns add input to each request.
- You register custom tools. Custom-tool configuration can route a request to a more capable model, even when it edits a built-in text block.
- Assistant image requests include conversation work. The image figures cover the image operation. Asking for an image in the Assistant also uses some text credits for the conversation.
- The image model sets the image cost. The default image models depend on the editor version, and the public API keeps its own default. A model you choose in your configuration can cost more or less than the defaults. The estimator marks image figures that predate a change of default models.
The figures exclude legacy AI tools such as Smart Text. Unlayer refreshes the figures when models, routing, or rates change; the dates under the estimate show when they were measured.
Check the estimate with your own usage
Before you commit to a budget, run representative requests in a test project
and compare them with the estimate. The
AI Credits usage API reports credits by end
user and feature_type, using the same features as this estimator:
| Estimator input | feature_type |
|---|---|
| Whole-design requests | full_template_gen |
| Text and block edits | block_edit |
| Images generated | image_generation |
| Images edited | image_edit |
| HTML imports | html_import |
| Screenshot imports | image_import |
To turn the estimate into per-user budgets that your app enforces, see End-User AI Credits & Upsell.