Koutuxiaremove.bg API Shutdown Dec 1: Leonardo.Ai Migration & Alternatives
Koutuxia does not provide an API service. This guide analyzes technical differences, pricing, and breaking changes when migrating from the remove.bg API to Leonardo.Ai or self-hosted alternatives.
From 1 December 2026, api.remove.bg/v1.0/removebg will stop accepting self-service requests. The official successor is the Leonardo.Ai API within Canva. This is not a simple hostname update: headers, payload schema, image encoding, response handling, and billing structures have all changed. Here is a technical breakdown of migration requirements, open-source options, and alternatives.
Key technical breaking changes in Leonardo.Ai
Payload protocol switched from Form-data to JSON: The legacy API accepted binary streams and multipart forms, whereas Leonardo requires a strict JSON payload where images are passed as Base64 strings or publicly accessible URLs.
Response handling no longer returns raw image streams: The legacy API streamed back the PNG binary directly. Leonardo returns a JSON response containing results[0].url by default, requiring a follow-up download step. Direct Base64 returns require setting "base64": true in the initial request.
Image retention and privacy behavior: Leonardo automatically archives outputs into your personal generation feed. To preserve privacy and prevent cloud storage accumulation, you must explicitly include "ephemeral": true, which invalidates the result link after 30 minutes.
Cost considerations: USD 0.1047 per image
Leonardo.Ai's migration guide cites a price of USD 0.1047 per image. For customers previously on high-tier monthly packages or annual discounted bundles, unit costs may see a noticeable increase.
Engineering teams running high-volume automated processing pipelines should recalculate projected monthly API expenditures before cutover.
Self-hosted alternatives: RMBG-2.0 and BiRefNet
For teams with available GPU infrastructure, state-of-the-art open-source models deliver production-grade matting accuracy. Bria-AI's RMBG-2.0 and BiRefNet natively support 1024x1024 inference resolution with fine hair and semi-transparent alpha preservation.
Deployed with ONNX Runtime or TensorRT behind FastAPI or gRPC, inference latencies can reach 100~300ms, keeping all proprietary media on internal networks while drastically reducing marginal cloud costs.
Non-automated workflows: Koutuxia local batch ZIP
Many smaller design and e-commerce teams initially built scripts simply to process 30 to 50 images per day rather than integrating deep microservices.
For manual recurring batches, Koutuxia's batch mode is a free, zero-code alternative: drop up to 50 images into the browser, let local hardware process them, and download a transparent PNG ZIP archive with zero API fees.
API alternatives and self-hosted matrix
| Option | Unit cost | Protocol | Data retention risk | Best suited for |
|---|---|---|---|---|
| Leonardo.Ai API (official) | USD 0.1047 / image | REST (JSON / Bearer) | Stored by default (requires ephemeral: true) | Existing automated pipelines tied to Canva |
| Photoroom API | USD 0.02 ~ 0.10 | REST (Multipart/JSON) | Optional zero-retention | E-commerce product cutouts at scale |
| Self-hosted RMBG-2.0 / BiRefNet | Hardware cost only | Custom (FastAPI/gRPC) | 100% on-premises, zero upload | Teams with strict data privacy and GPU hardware |
| Koutuxia batch mode | Free (100% local) | Web UI (no API) | Runs in browser RAM, zero server upload | Manual batches up to 50 photos at a time |
Leonardo.Ai pricing from official migration guide. Photoroom pricing from public pricing page (checked September 2026).
Developer migration checklist
Step 1 · Audit monthly call volume and budget
Examine past API usage logs and model expenses against Leonardo's USD 0.1047 rate to confirm economic viability.
Step 2 · Refactor request and response handling
Update endpoints, migrate authentication to Bearer tokens, adjust payloads to JSON, and implement result URL fetching or Base64 decoding.
Step 3 · Enable ephemeral mode
Add "ephemeral": true to API parameters in production to ensure processed client images are not permanently retained in cloud storage.
Who this guide is for
- ·Software engineers and backend teams maintaining remove.bg API integrations
- ·Technical leads evaluating self-hosted AI models vs commercial API services
- ·Small teams looking for a free, no-code local batch alternative
FAQ
Can I reuse my existing remove.bg API key on Leonardo.Ai?
No. You must create an account on Leonardo.Ai and generate a separate API key passed as a Bearer token in the Authorization header. (Source: Leonardo.Ai docs)
Does the shutdown affect enterprise remove.bg API contracts?
No. Enterprise API agreements remain active under their existing contract terms and are not terminated on 1 December. (Source: remove.bg FAQ)
Does Koutuxia offer a public REST API?
No. Koutuxia runs exclusively as a client-side in-browser web tool and has no plan to launch a commercial public API before 1 December.
Sources
- Leonardo.Ai: Migrate from the remove.bg APIdocs.leonardo.ai
- Leonardo.Ai: Remove bg endpoint referencedocs.leonardo.ai
- remove.bg official API shutdown noticewww.remove.bg
- Photoroom API pricing and docswww.photoroom.com
- Bria-AI RMBG-2.0 on Hugging Facehuggingface.co
- Migrating from remove.bg: steps for four kinds of user
No code needed for batches: run in browser
Process up to 50 photos locally and download a zipped package of transparent PNGs.
Remove background free