Amazon Nova Canvas alternative: edit and generate images on Sume

Nova Canvas works through Amazon Bedrock with task types: text-to-image, inpainting, outpainting, variation, background removal, color guidance and try-on (Amazon docs, read 2026-10-06). Teams look for an alternative when they want one HTTP call and a wider model choice. Sume's Image API is a single endpoint, POST /v1/images, with a catalog of models behind it.

What an edit costs on Sume

Billed price is the list price times 1.25. For an edit-capable model, the Ideogram 4.5 medium edit lists at $0.06 and bills $0.075; a Nano Banana 2 image lists at $0.08 and bills $0.10; a GPT Image 2.5 image lists at about $0.0527 and bills about $0.066. Read the exact numbers from GET /v1/images/models/{id}/endpoints rather than from a blog.

Try one task end to end

The call below is the smallest swap for an inpainting step. Run it on a few real photos before you decide.

import os, requests

r = requests.post("https://api.sume.com/v1/images", timeout=90,
    headers={"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"},
    json={"model": "sume/auto",
          "prompt": "Remove the person on the left. Fill with the background.",
          "input_references": [{"type": "image_url", "image_url": {"url": "https://example.com/street.jpg"}}]})
print(r.status_code)
body = r.json()
print(body.get("usage") or body)

Where Nova Canvas gives you a named task and a mask, Sume gives you a prompt and a choice of model. If you rely on exact mask control, test mask_url first.