Cover: AI-generated editorial illustration.
Evergreen guide. Prices and details checked on 8 November 2026 against the linked pages; prices can change, so check them before you buy.
The remove.bg API is moving to Leonardo.Ai on 1 December 2026, as stated on the remove.bg API page. If your code calls the remove.bg endpoint today, you have until that date to test and deploy an alternative. Opennan's background removal API costs $0.0234 per image with no subscription required and failed runs not charged.
What is changing at remove.bg
The remove.bg API page says: "Starting December 1, 2026, background removal moves to Leonardo.Ai, also part of Canva. Same simple integration, plus a full creative toolkit that picks up right where background removal leaves off." The page describes Leonardo.Ai as adding capabilities beyond background removal: automated upscaling, AI-generated scenes, image animation, and multi-step workflows it calls Blueprints. Even if the core integration stays simple, the platform itself is broader and different, so it is worth verifying that your specific request format, response handling, and pricing expectations still match before the transition date.
Background removal API prices
The table below includes only providers whose current public pricing pages show a specific per-image price for background removal. Providers not found in the evidence are not listed.
| Provider | Price per image | Plan conditions |
|---|---|---|
| Bria | $0.018 | Pay-as-you-go; $0.025 with prompt. Source: bria.ai/pricing |
| Photoroom | $0.02 | Basic plan (Remove Background API); monthly subscription required. Source: photoroom.com/api/pricing |
| Opennan | $0.0234 | Pay-as-you-go, no subscription; failed runs not charged. Source: opennan.com/products/image-remove-background |
Bria's pay-as-you-go rate is the lowest listed here. Photoroom requires subscribing to a monthly plan before you can use production API calls beyond the initial 10 free ones. Opennan charges per use with no plan or commitment, and the same API key and balance work across all its other tools.
Switching from remove.bg to Opennan
The remove.bg API uses a multipart form POST to https://api.remove.bg/v1.0/removebg with an X-Api-Key header. Opennan uses a JSON POST to https://api.opennan.com/v1/images/background-removals with a Bearer token header.
Your existing remove.bg call in Python looks like this:
requests.post('https://api.remove.bg/v1.0/removebg', data={'image_url': url, 'size': 'auto'}, headers={'X-Api-Key': KEY})
The equivalent Opennan call:
r = requests.post('https://api.opennan.com/v1/images/background-removals', json={'image_url': url}, headers={'Authorization': f'Bearer {OPENNAN_KEY}'})
Then poll the returned task until it finishes:
requests.get(f"https://api.opennan.com/v1/tasks/{r.json()['id']}", headers={'Authorization': f'Bearer {OPENNAN_KEY}'})
Three things are different from remove.bg. First, the call is asynchronous: Opennan returns a task ID immediately and you poll GET /v1/tasks/{id} until the status reaches succeeded or failed. Second, authentication uses a standard Bearer token instead of a custom header. Third, the request body is JSON rather than multipart form data. The Opennan API docs show the full task object, all status values, and error codes with suggested fixes.
Honest reasons to consider each option
Bria lists the lowest per-image price at $0.018 on its pay-as-you-go plan. Its pricing page notes that pay-as-you-go includes capped standard indemnification, while its Enterprise plan offers unlimited IP indemnification backed by fully-licensed training data. If IP protection for commercial output is a priority, that distinction matters.
Photoroom's Basic plan at $0.02 per image covers background removal alongside crop, resize, and colour background features. The pricing page shows 10 free production calls on account creation and 1,000 free sandbox calls per month on the Image Editing API. A monthly subscription is required to continue beyond the free allocation.
Opennan at $0.0234 per image is pay-as-you-go with no subscription. The product page states that failed runs are not charged, the website lets you test with 3 free removals per day without an account, and signing in free gives you the full-quality PNG download. The API accepts PNG, JPG, or WEBP up to 20 MB and 25 megapixels. Because one key covers all Opennan tools on a shared balance, teams already using Opennan for other tasks pay from the same account without managing separate credentials.
FAQ
Is remove.bg shutting down on 1 December 2026?
The remove.bg API page says the service "moves to Leonardo.Ai" on that date. It does not use the word shutdown. The page describes it as moving to a new platform with an extended feature set. What that means for pricing and any edge cases in your existing integration is not fully detailed on the current page, so testing before the date is advisable.
Does Opennan charge for a failed background removal run?
No. The Opennan product page states explicitly that failed runs are not charged. The API docs confirm this for tasks generally: a failed task carries "refunded": true on the error object.
What file formats and sizes does Opennan accept?
PNG, JPG, or WEBP, up to 20 MB and 25 megapixels, as listed on the Opennan product page. The result is a transparent PNG the same size as the input.
Can I evaluate quality before paying?
Yes. The Opennan website offers 3 free removals per day without an account. After a free sign-in you can download the full-quality PNG. This lets you check output on your own images before adding any API credit.
Can an AI agent call the Opennan background remover?
Yes. The product page shows that adding the MCP server at https://api.opennan.com/v1/mcp exposes a remove_image_background tool that AI agents can call directly with an image_url argument.
If you need a pay-as-you-go background removal API before the remove.bg transition date, the Opennan background removal page has the full API reference, code samples for curl, Python, and JavaScript, and a free web tool where you can test your own images before writing a line of code.