Is this the same API used in production?

Yes. The demo calls the identical /v1.0/image-without-background endpoint. No toy model, no sampling differences.

What is the actual API latency?

Server-side processing: p95 ≈ 800 ms in Frankfurt, DE on AWS Inferentia. End-to-end time adds your upload + TLS handshake + queue + download.

Do you store or train on my images?

No. Images are processed in RAM and discarded immediately after the response. No disk writes, thumbnails, caches, or backups. Customer uploads are never used for model training.

What are the rate limits?

The API is credit-based with no subscription lock-in. You start with 50 free credits. See the documentation for current rate limit details.

Can I self-host instead of using the cloud API?

Yes. The same model family is available as an Apache-2.0 open-source package: pip install withoutbg, or run the one-command Docker image. No outbound data.

withoutBG

Background removal API, test it live

Same /v1.0/image-without-background endpoint as production. No signup to demo. Upload any image and see the result plus the latency and the exact curl command to reproduce it.

Two models
/api-model · /open-weights-model
Precision cutouts
Hair, Fur, Glass, Motion
Run your way
self-host · API · plugins
Explore
/docs · /tech · /compare

Get API key API docs Sample results

No signup required to demo. Self-host option available under Apache-2.0. 1,158 stars

Quality on hard cases

See 50+ examples

Close-up street portrait with windblown curly hair and dramatic backlight, challenging fine hair edges

AI-generated with midjourney

Original image: Windblown Curly Portrait — Original Image

Background removed: Windblown Curly Portrait — Without Background

Alpha matte prediction: Windblown Curly Portrait — Alpha Matte (Model Prediction)

Small balcony plant with bokeh background and thin leaves and stems

Shot with Canon 5D (Canon 5D Mark II with Takumar 50mm f/1.4)

Original image: Balcony Plant Bokeh — Original Image

Background removed: Balcony Plant Bokeh — Without Background

Alpha matte prediction: Balcony Plant Bokeh — Alpha Matte (Model Prediction)

Request / Response

JSON base64 endpoint. A multipart binary endpoint is also available, see docs.

Request bodyPOST /v1.0/image-without-background-base64

{
  "image_base64": "/9j/4AAQSkZJRgABAQAA…"
}

Response body

{
  "img_without_background_base64": "iiabVBORGgoAAAANSUhEUgAAASw...."
}

The demo shows latency in the UI after each call, but it is not part of the published base64 response schema.

Production-grade, privacy-first

Region: Frankfurt, DE
EU data residency. No cross-border data transfers.
RAM-only processing
Images never written to disk, thumbnails, caches, or backups.
No training on customer data
Uploads are never used to fine-tune or retrain models.
TLS in transit
All requests over HTTPS. API keys transmitted as headers only.
AWS Inferentia · p95 ≈ 800 ms
Server-side processing. Total time = upload + TLS + queue + ~800 ms + download.
Open-source alternative
Apache-2.0. Same model family runs locally via pip or Docker, zero outbound data.

Prefer to keep images on-premise?

The same model family runs entirely offline under Apache-2.0. No API key, no outbound data.

pip install withoutbg

docker run -p 80:80 withoutbg/app:latest

github.com/withoutbg/withoutbg1,158 stars

Common questions

Is this the same API used in production?: Yes. The demo calls the identical /v1.0/image-without-background endpoint with no differences in model or sampling.
What is the actual latency I will see?: Server-side processing is p95 ≈ 800 ms in Frankfurt, DE on AWS Inferentia. End-to-end time also includes your upload, TLS handshake, queue, and download. The demo shows your actual measured latency after each call.
Do you store or train on my images?: No. Images are processed in RAM and discarded immediately after the response. No disk writes, thumbnails, caches, or backups. Customer uploads are never used for model training or fine-tuning.
What are the rate limits?: The API is credit-based with no subscription lock-in. You get 50 free credits to start. See the documentation for the current per-key concurrency and burst limits.
Can I self-host instead of using the cloud API?: Yes. The same model family is available open-source under Apache-2.0: pip install withoutbg runs local inference, or docker run -p 80:80 withoutbg/app:latest spins up a full web UI with zero outbound data.

Ready to integrate?

Get 50 free API credits, no credit card required. Start removing backgrounds in minutes.

Get an API key and start implementationGet 50 free API credits.