Bedrock image features Sume does not have: an honest gap list
Moving off Bedrock image models? These Nova Canvas and Titan features have no Sume equivalent: maskPrompt, mergeStyle, similarityStrength, fine-tuning.

Before you migrate, know what you will not find. This is the list of Bedrock image features with no field on Sume's /v1/images, taken from Amazon's pages (read 2026-10-06) and Sume's docs. Sume's API returns 400 unsupported_parameter for any field a model's catalog row does not list, so these fail loudly rather than being ignored.
| Bedrock feature | What it does | Workaround on Sume |
|---|---|---|
| maskPrompt | Names the region to edit in text | Put the region in prompt |
| mergeStyle (try-on) | BALANCED, SEAMLESS or DETAILED blending | Two references and a prompt; measure the seam |
| similarityStrength | 0.2 to 1.0 closeness to the input | Wording; pick a model |
| outPaintingMode | DEFAULT or PRECISE | Set aspect_ratio and diff the middle |
| cfgScale | 1.1 to 10 prompt adherence | quality where listed |
| Fine-tuning (Titan) | Train on your own images | References on every call |
| Watermark detection API (Titan, Nova) | Checks for the Amazon invisible watermark | None |
What Sume has that Bedrock does not
Sume lists many image models behind one schema, so you can run the same edit through several and compare. It also returns a job envelope for slow requests and a usage.cost on every response. Failed jobs are not billed.
Find out what a model accepts, per model
Run this once in CI so a port does not break on a new model. It lists the models that take input_references, which is the shortlist for edits.
import os, requests
r = requests.get("https://api.sume.com/v1/images/models", timeout=30,
headers={"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"})
r.raise_for_status()
for m in r.json()["data"]:
refs = m["supported_parameters"].get("input_references", {})
if refs.get("max", 0) > 0:
print(m["id"], "refs up to", refs["max"])Sources
Related posts
More in Comparisons
- ffmpeg script or Sume Timeline for a weekly vertical series
Compare a self-hosted ffmpeg script with Sume Timeline 1.0 for a weekly 9:16 series: what you maintain, what the API refuses, price per output minute.
- Groq Orpheus TTS: $22 per million characters and a 200-character cap
Groq lists Orpheus V1 English at $22.00 per million characters with input kept under 200 characters. Request-count and cost math against Sume TTS.
- Ideogram 4.5 is 0.8c to 22c per image: which tiers does Sume sell?
Ideogram lists four quality modes from 0.8 cents to 22 cents at native 2K. Sume's Image API sells low, medium and high at $0.0375, $0.075 and $0.275 an image.
- Lemonfox TTS at $2.50 per million characters vs Sume: the honest gap
Lemonfox's page works out to about $2.50 per million characters. Sume lists $47.50. Cost for 900, 22,500 and 1M characters, and what the gap buys you.
Written by Sume