If you run a background-removal pipeline on remove.bg, you'll need a new provider soon. On December 1, 2026, at 9:00 AM CET, remove.bg's self-service API stops accepting requests. remove.bg recommends two replacements, both owned by Canva: Canva itself for single images and bulk batches, and the Leonardo.Ai API for larger-volume and automated use. Neither switch happens automatically.

So the real question is what background removal will cost you once you move. For some teams, the Leonardo.Ai API is an easy call that saves money. For others, it's a price increase that gets steeper the more images they process.

The good news is that Leonardo.Ai isn't your only option. You have to set up a new account and rewrite your integration anyway, so this is the easiest time you'll ever have to switch providers. You won't be tearing out a working setup later. That makes now the right moment to compare Leonardo.Ai with other options like Photoroom's API.

Ultimately, the best choice comes down to how many images you process each month.

Who this guide is for: self-service API users. If you only use the remove.bg website, remove.bg recommends Canva. If you're on an enterprise API contract, your contract continues as normal. remove.bg's plugins and integrations also end on December 1, 2026, so if your pipeline runs through one, you'll need to replace it too.

Table of contents

What does the remove.bg to Leonardo.AI migration mean for API pricing?

Remove.bg sells credits, with one image requiring one credit to remove the background. In addition to offering a pay-as-you-go option, ranging from three credits for $3 to 8,000 credits for $1,699, remove.bg offers multiple subscription plan tiers, starting at 40 credits for $9/month and scaling to 75,000 credits for $5,450/month.

Leonardo prices background removal at a flat $0.1047 per image. That makes the bill easy to forecast and also means there’s a crossover point, and which side of it you’re on decides whether the remove.bg API migration saves you money or costs you more.

Because remove.bg's per-image rate falls as the tier size rises while Leonardo's stays flat, the two lines cross. Up to the 7,500-image tier ($789/month, or $0.1052/image), Leonardo.Ai’s flat rate is lower than remove.bg’s subscription pricing on a per-image basis. Above it, remove.bg’s per-image rate continues dropping and Leonardo’s doesn’t, meaning you will pay more for each image at higher volumes.

These figures compare self-service plans only, and they assume you choose Leonardo.Ai. At the smallest tiers, remove.bg recommends Canva for manual use, so the savings shown apply only if you call the API from your own code. At the top tiers, enterprise customers keep their existing contract pricing.

In short, customers processing fewer images each month stand to save money. For small operations, they may see their bill cut roughly in half. Teams at the high end will pay more: at 75,000 images a month, the Leonardo.Ai API costs about $2,400 more per month than remove.bg did. The bigger your catalog, the more this migration costs you.

How does Photoroom compare?

Photoroom’s Remove Background API is part of the Basic plan, which is priced at $0.02 per image. With this plan, you’ll also be able to crop or resize images, add color backgrounds, apply green screen despill, and get uncertainty scoring.

Thanks to its low flat rate, Photoroom costs less than both remove.bg and Leonardo.Ai from about 200 images a month upward. Here's how it stacks up.

So whether you process a few hundred images a month or tens of thousands, Photoroom’s rate is lower than every remove.bg tier and lower than Leonardo.Ai’s flat rate.

It’s also worth noting that both Leonardo and Photoroom will price by negotiation at volume. Leonardo offers a custom API plan with model-based discounts for those with large workloads, while Photoroom offers scalable pricing with volume-based discounts through the Enterprise plan for those able to commit to 200K+ images per year.

Do credits and images expire?

Another thing to check as you leave remove.bg is what happens to the images you pay for but don't use.

The published per-image rates in the table above assume you use every image you pay for. Your real cost per image is what you paid divided by the images you processed, and expiry rules can impact your cost per image.

Take remove.bg's 500-image plan at $89 a month. That's $0.178 per image on paper. In a month when you process only 300 images, you still pay $89, so you actually paid about $0.297 per image. The same logic applies to any plan where unused volume disappears at the end of the period.

On the other hand, in plans where unused volume carries over, you will likely eventually use what you paid for. As a result, your real rate stays close to the published one.

Here’s a look at how remove.bg, Leonardo, and Photoroom handle unused credits or images:

Expiry matters most if your volume is uneven, whether from seasonal peaks, product launches, or catalog refreshes. But the starting rate matters more. Photoroom's images don't roll over, yet at $0.02 per image you'd have to leave more than 80% of them unused before your real cost reached Leonardo.Ai's $0.1047. In the same slow month as the remove.bg example above, using 300 images on Photoroom's 1,000-image plan ($20), you'd pay about $0.067 per image, still well below Leonardo.Ai's $0.1047 and remove.bg's $0.297.

Leonardo.Ai API and Photoroom API coding requirements

But cost isn’t all about money. Because you have to rewrite the request either way, it's also worth looking at what each path asks of your code.

For reference, the remove.bg API is an HTTP interface you call with a simple request. Here's a cURL example:

curl -H 'X-API-Key: INSERT_YOUR_API_KEY_HERE'           \

       -F 'image_file=@/path/to/file.jpg'                 \

       -F 'size=auto'                                     \

       -f https://api.remove.bg/v1.0/removebg -o no-bg.png

With remove.bg, it’s possible to receive 10MP PNGs, 50MP JPGs, 50MP WebPs, and 50MP ZIPs. The ZIP format is faster for transparent images, but JPG is recommended if you don’t need transparency.

Leonardo.Ai

Leonardo.Ai has two ways to call the remove.bg model:

• Sync API: This is ideal for single images and will return the processed image, no polling required. The service will time out after approximately 27 seconds.

• Async API: Here, you submit a job and retrieve the result once it completes, instead of waiting on an open connection. As a result, it’s better suited for batch processing, very large images, workloads that take longer than 27 seconds, or pipelines that shouldn’t keep the connection open.

The Sync API returns JSON with a link to the processed image, while the Async API returns a generation ID immediately. You can then use this ID to retrieve the final image.

Things that don't carry over from remove.bg include file uploads with multipart/form-data (requests are JSON only, with images sent as base64, a URL, or an uploaded image ID, up to 25MB), ZIP output, format=auto, attaching a background image file, raw image bytes in the response, and foreground position metadata in X-Type / X-Foreground headers.

Sync API cURL example

curl --location 'https://cloud.leonardo.ai/api/rest/v2/generationssync' \

     --header 'authorization: Bearer {api-key}' \

     --header 'Content-Type: application/json' \

     --data '{

       "model": "remove-bg",

       "public": false,

       "ephemeral": true,

       "parameters": {

         "size": "auto",

         "type": "auto",

         "format": "png",

         "guidances": {

           "image_reference": [

             {

               "image": {

                 "id": "YOUR_IMAGE_ID",

                 "type": "UPLOADED"

               }

             }

           ]

         }

       }

     }'

Sync API JSON response example

{

  "id": "b3a6c1d2-…",

  "blockedCount": 0,

  "cost": { "amount": "0.1047", "unit": "DOLLARS" },

  "results": [

    {

      "url": "https://…",

      "contentType": "image/png",

      "width": 1024,

      "height": 768

    }

  ]

}

Async API cURL example

curl --location 'https://cloud.leonardo.ai/api/rest/v2/generations' \

     --header 'authorization: Bearer {api-key}' \

     --header 'Content-Type: application/json' \

     --data '{

       "model": "remove-bg",

       "public": false,

       "parameters": {

         "size": "auto",

         "type": "auto",

         "format": "png",

         "guidances": {

           "image_reference": [

             {

               "image": {

                 "id": "YOUR_IMAGE_ID",

                 "type": "UPLOADED"

               }

             }

           ]

         }

       }

     }'

Async API JSON response example

{

  "generationId": "b3a6c1d2-…",

  "apiCreditCost": 1

}

To get the result, call "GET https://cloud.leonardo.ai/api/rest/v2/generations/{generationId}" with your API key, or set up a webhook to receive it automatically.

Photoroom

If you want to call the Remove Background API in Photoroom, you can run this cURL code:

curl --request POST \

  --url https://sdk.photoroom.com/v1/segment \

  --header 'x-api-key: YOUR_API_KEY' \

  --form image_file=@/path/to/your/image.jpg \

  --output photoroom-result.png

This removes the background and saves the result as a PNG.

Additionally, since Photoroom’s API takes the same argument names as remove.bg, migrating is fast and easy. All you'll need to do is change your endpoint URL and your key. Your main parameters, including image_file, crop, bg_color, format, and size, carry over unchanged. It also accepts the same multipart file upload, so your upload code doesn't change.

How to evaluate the right background removal API for your pipeline

Since staying with remove.bg isn’t an option, now is a good time to decide what you want from a background removal API.

1. Start with your monthly image volume

Your volume decides the price, but only if you count it correctly. You’ll need to consider:

• Your slowest month: Plans where unused volume disappears cost more per image in slow months. This is the case with most providers, so if you often have slow months per-image cost becomes all the more important.

• Your busiest month: Leonardo.Ai and Photoroom both charge flat rates, so your bill grows in a straight line. At 50,000 images, that's about $5,235 on Leonardo.Ai ($0.1047 per image) and $1,000 on Photoroom ($0.02 per image, on every plan from 1,000 images). remove.bg worked differently: its per-image rate fell as you moved up tiers, from $0.225 at 40 images to $0.0727 at 75,000. That's why Leonardo.Ai costs less than remove.bg up to about 7,500 images a month and more above it, with the gap growing as you add images. Photoroom costs less than both from about 200 images a month.

• Your yearly total: If you process a large number of images each year, custom pricing may bring your per-image cost below the published flat rate. Leonardo offers a custom API plan with model-based discounts, and Photoroom's Enterprise plan offers volume-based discounts for teams that can commit to 200,000 or more images per year.

2. Look at the rest of your image pipeline

Background removal is usually one step in a longer chain, so look at what happens before and after it. Photoroom's Basic plan also covers cropping and resizing, adding color backgrounds, green screen despill, and uncertainty scoring. If your pipeline does any of that in separate steps, one API may replace several.

Then check what your code depends on in remove.bg's responses. Leonardo.Ai doesn't carry over ZIP output, attaching a background image file, raw image bytes in the response, or the foreground position metadata in the X-Type and X-Foreground headers. If your code reads any of those, budget rewrite time for it. The biggest change for most teams is the upload itself: Leonardo.Ai doesn't accept multipart file uploads, so if your code attaches files the way remove.bg's cURL example does, you'll need to switch to base64, a URL, or an uploaded image ID.

Integration shape matters too. Leonardo.Ai offers a sync API that returns JSON with a link to the processed image and times out after about 27 seconds, and an async API that returns a generation ID you use to retrieve the result later. The async route suits batch jobs and very large images. Meanwhile, Photoroom’s Remove Background call is a single POST that returns the image.

3. Test your difficult images

Rates and features don’t show how a service handles your hardest images, which is one of the most important factors when choosing a background removal API.

Take the 20 that have given you the most trouble (think fine hair, glass, translucent fabric, low-contrast edges, and cluttered backgrounds). Then, run the same set through each API and compare the cutouts side by side. Price the result, including any retries or manual cleanup. 

Photoroom’s uncertainty score can help here by flagging results that need a second look. It returns a number between 0 and 1, with numbers closer to 0 meaning the model is quite confident that the cutout is accurate. Scores closer to 1 mean the model is more unsure.

For a head start before you run your own test, see how Photoroom and remove.bg compare on product images.

Picking your next background removal API

With remove.bg wrapping up on December 1, 2026, now is the perfect time to take a step back and really think about what you want from your next background removal API, since you have to rewrite your integration either way.

Leonardo.Ai’s flat-rate pricing structure means a predictable $0.1047 per image at any volume. That's a saving if you process fewer than roughly 7,500 images a month, and a growing increase if you process more. On the other hand, Photoroom’s Basic plan at $0.02 per image has a lower cost per image than every remove.bg tier from 200 images up and is about a fifth of Leonardo’s flat rate.

Price is only part of the switch, so think holistically and weigh what each API asks of your code, what it drops, and how it handles your volume before committing to a new background removal API.

*Photoroom's smallest Basic plan includes 1,000 images for $20 a month.