The true cost of AI-generated product cards: calculating actual results, not just advertised "starting from" prices

When creating a card for a real product, a single slide might seem inexpensive. However, a marketplace product card involves more than just one slide or one generation attempt.

It includes source images, multiple scenes, infographics, specific product placement requirements, text readability, detail quality, and, crucially, human review. Even a high-quality AI tool for product cards cannot replace the need for visual inspection of the final result.

Therefore, it makes more sense to calculate the cost of a product card based on the price of a complete, ready-to-use set rather than the cost of a single generation.

How we calculate the cost of one product card

Let’s take a specific task. A designer creates a product card for a women’s linen suit for online marketplaces. The card consists of five slides:

1. cover image with the model and suit;

2. full view of the suit;

3. close-up of the fabric and details;

4. slide highlighting the model’s benefits;

5. final slide with specifications.

We use Aidentika — a service specializing in visual content for products. To create a product card, simply upload a product photo, choose a concept, and write key points for the infographic. Multiple images can be created in a consistent visual style.

If you only look at the tariffs and the cost of a single generation, you might think that five slides cost next to nothing. However, it is important to first understand exactly what sparks are deducted for.

How much does one generation cost in the neural network for cards

In Aidentika the internal currency is called sparks. According to current service information, generating a photo or card in 1K/2K resolution costs 4 sparks. Enhancing an already created card costs an additional 2 sparks. For 4K, the generation cost is 6 sparks.

This is an important clarification because the figure “2 sparks” does appear in Aidentika’s tariff logic, but it refers to enhancement, not basic card generation.

The website offers four packages:

The site also provides a guideline for the number of generations: 

• 50 sparks are enough for approximately 12 images, 

• 140 — for 35, 

• 400 — for 100, 

• 1200 — for 300. 

This means the actual cost per basic generation decreases as the package size increases. This allows you to calculate the real unit cost rather than relying on marketing "starting from" prices.

Creating a product card with AI: calculating for five slides

Our product card requires five images. One generation costs 4 sparks.

So: 5 slides × 4 sparks = 20 sparks.

If you buy the minimum "Start" package, where 50 sparks cost $5, one spark costs about $0.10. Then five slides: 20 × $0.10 = $2. It turns out that with the minimum package, generating five slides costs about $2.

But this is not yet the price of a finished card. Because we currently assume that all five generations were immediately suitable for publication. In real work, a designer opens each result and checks it. And here repeated deductions appear.

Where AI product card generation might require another attempt

AI can create a visually beautiful image and simultaneously make a mistake in a detail that matters for a product card.

For example, buttons or other small elements on a suit may differ from the original. Text on an infographic slide may require verification.

During the first generation attempt for the fifth slide, the suit had completely different buttons.

Regeneration using a reference with the correct button turned out well.

This is precisely why a good generation result cannot be evaluated solely on the principle that "the picture looks beautiful."

There is also the problem of preserving product characteristics: neural networks can convincingly show clothing on a model, but individual details, fit, or garment elements may become distorted. For a seller, this is not just a cosmetic flaw, but a reason to check the image before publication.

Therefore, let's include three additional generations in the calculation.

This does not mean that every slide will necessarily have to be redone three times. This is a conservative scenario that shows the economics of a card if some results need to be replaced.

1. 3 repeat generations × 4 sparks = 12 sparks.

2. In the minimum package, this is another: 12 × $0.10 = $1.20.

3. Now the cost of generations is: $2 + $1.20 = $3.20.

And this is still not the final amount.

AI-generated product cards still require a designer’s review

This is where misconceptions often arise. One might ask: “If AI has already created the infographics, positioned the product correctly, and delivered high quality, why is a designer needed at all?”

The designer then opens the completed set and checks each card. They verify whether the product matches the source image, ensure no extra elements slipped into the frame, check that text isn’t too small, confirm that infographics don’t obscure important parts of the item, ensure consistency across all five slides, and look for technical errors. Even if nothing needs to be completely redone, time is still required for verification and minor adjustments.

For our calculation, we’ll allocate about $4 for the designer’s time spent on manual review and refinement of the set. In this case, manual editing was needed for the second slide because the AI duplicated the suit color range, even showing different textures. We need to manually remove one of the lists:

Here, the classic approach works best: open Photoshop or Figma, erase the duplicate, and clean up the background. It takes 15 minutes, resulting in a good slide without lengthy explanations to the AI about what it should have done:

$4 is not Aidentika’s rate nor a statement about fixed designer costs. It is simply a calculated scenario that helps illustrate the full economics.

The result is:

$3.10 for generation + $4 for the designer’s work = $7.10. Rounded, that’s about $7 for a finished five-slide card, based on the minimum package and three additional generations.

What if you choose a larger package?

The economics change here. The larger the package, the lower the cost per credit. For example: 

1. In the “Studio” package, 400 credits cost $31. That’s approximately $0.08 per credit. 

2. Then the initial five slides: 20 × $0.08 = $1.60.

3. Three additional generations: 12 × $0.08 = $0.96.

4. Total for generations: $2.56.

5. Adding the same $4 for manual work: $6.56.

So, with higher order volumes, one five-slide card in our scenario costs about $7, including three additional generations and the designer’s work.

This is why package pricing matters for sellers managing dozens of SKUs rather than just one card. At such volumes, even a small difference in the cost per generation significantly impacts the budget.

After registration, your first turnkey card is free

Why “starting at $0.50 per card” is not the best way to calculate costs

Let’s return to the figure we started with. If we estimate a slide at roughly $0.10, five slides will cost: 5 × $0.10 = $0.50.

Add three regeneration attempts: 3 × $0.05 = $0.15.

The total comes to $0.65.

On an advertising banner, this might look like the cost of a single card. But designers understand that $0.65 covers only a set of generation attempts, not the final result. If we add about $4 for labor time:

52 + 300 = 352 ₽.

This is the figure to keep in mind when comparing tools: even with very optimistic calculations, a card is no longer just a “$0.50 card.”

Note that this calculation uses an estimated four generation credits per initial slide. Therefore, for realistic budget planning, it is better to rely on the higher estimate—around 6–7 in our scenario, depending on the chosen package.

Aidentika card generator: where does the savings boundary lie?

The main value of Aidentika here is not that it completely eliminates the designer’s work. It changes the volume of their workload.

Instead of building five slides from scratch, the designer receives pre-generated images and focuses on quality control. The service claims automatic infographic creation: the user uploads photos, selects the Card tool, and enters text with product specifications. You can also create multiple variants and choose the most suitable one.

This is especially important for product series. When you need to make just one SKU, the savings may seem modest. But when there are ten, twenty, or fifty such items, reducing manual work begins to impact the entire production process.

However, you cannot exclude regeneration attempts and verification from the calculation. The AI does not know which details are critical for a specific brand until they are specified and checked by a human.

Therefore, it is more accurate to evaluate Aidentika not by the cost of a single button click, but by the price of a card that is truly ready to be published on marketplaces.

How much does it cost to create a product card: final calculation

Let’s compile everything into a single table.

This is why the question “how much does one product card cost” cannot be answered with a single figure from an advertisement.

For a single card, you need to account for the number of slides, the cost of each generation, retries, the size of the spark package, and specialist time. Some users get the desired result on the first try, while others require several replacements. One designer may only check the finished images, while another will adjust the composition, text, or specific product details. Therefore, AI tools for creating product cards should be evaluated not by the price of a single image, but by the cost of the entire production cycle.

If the goal is simply to get a nice picture, you can count the generation cost. If you need to create a product card using AI and list it on marketplaces, you have to calculate based on the final result. This is the key metric to keep in mind before comparing services next time: the generation price is not the card price.

Get 6 free sparks at start – enough for a slide + regeneration