Transparent sticker: GPT Image 2.5 background field or Nano Banana 2.1
Only GPT Image 2.5 lists a background field on Sume. Nano Banana 2.1 does not, so a transparent PNG there is $0.10 plus a $0.0225 background-removal step.

For a transparent sticker, ChatGPT Image 2.5 is the Sume image row that takes a background field (auto, transparent or opaque). Nano Banana 2.1 does not list that field, so a transparent PNG from it means a second step: $0.10 for the image at 1K plus $0.0225 for background removal, $0.1225 in all.
What the catalog lists
The Sume docs say ChatGPT Image 2.5 supports background, and the catalog code adds it only to the two GPT Image 2.5 rows. A request that sets a parameter a model does not list returns 400 unsupported_parameter rather than ignoring it.
| Path | Calls | Arithmetic | Cost per sticker |
|---|---|---|---|
| openai/gpt-image-2.5 with background transparent | 1 | set quality and size, read the quote | varies |
| google/nano-banana-2.1 at 1K plus background removal | 2 | 0.10 + 0.0225 | $0.1225 |
| google/nano-banana-2.1 at 0.5K plus background removal | 2 | 0.075 + 0.0225 | $0.0975 |
Which to use
If you make hundreds of stickers, price both. The GPT row costs depend on quality and size, so quote one sample before multiplying. The two-step path has a fixed price you can compute up front, but it adds a second job to track and the cut-out edge can be rough on hair or glass.
Prompt tip
Ask for a flat, single-color or white background and a clear outline when you plan to cut it out later, and keep shadows off the ground plane. A clean cut starts with a clean source.
Checking a cut-out
Put each result on a dark and a light background and look at the edge. Halos show up on one of them. If the edge fails, regenerate with a plainer background rather than fixing it by hand, because at $0.10 or less a retry is cheaper than editing time.
For GPT Image 2.5, request output_format: "png" when the background is transparent. Sume lists png, jpeg and webp for the row, and a transparent image needs an alpha channel, which JPEG does not have.
Price a pilot of 20 stickers on each path before you commit a large run. The pilot costs about $2 to $3 and shows the failure rate.
Costs to remember
The GPT Image 2.5 path has no second step, but its price needs a quote because it is token-based. Request a single sticker with quality and image_size set, read usage.cost, then multiply. The two-step Nano Banana 2.1 path has a known price, $0.1225 at 1K, but two jobs.
Whichever path you take, store the original and the cut-out. If you later move to a different cut-out tool, you can redo the second step without paying for the image again.
For a small shop, a rough plan is 20 stickers a week. On the two-step path at 1K that is 20 x $0.1225 = $2.45 a week. At 0.5K it is 20 x $0.0975 = $1.95. A weekly total under $3 means the choice between paths is about quality and your time, not money. If you run 2,000 stickers, the same arithmetic gives $245.00 and $195.00, and the pilot is worth doing carefully.
Sources
Related posts
More in Media tools
- Trim a 4-minute video to a 180 s YouTube Short: start 30, duration 180
YouTube's Shorts page allows up to 3 minutes. Sume video trim cuts 180 seconds from a 240 second source for $0.02. The request, the limits and the refusals.
- Video trim caps: 1800 s in, 900 s out, and the 901 s refusal
Sume video-trim accepts a source up to 1800 seconds and an output up to 900 seconds. The refusal codes you get past either cap and how to split a long cut.
- video-inspect vs video-frames: which returns the still you need
Video inspect gives a probe, 8 default stills at 768 px and optional STT for clips up to 1800 s; video frames gives exact stills at source size up to 300 s.
- Sentence segments from video inspect: caption lines with no gaps
Sume video inspect with transcribe true can return gapless sentence segments, with silence_split_seconds from 0.2 to 3. Feed them to captions as cues.
Written by Sume