AI image for print: pixel sizes for 300 DPI prints

To use an AI image for print, make sure it has enough pixels for the print size: inches equal pixels divided by DPI, so a 300-DPI print needs 300 pixels for every inch of width and height. On Sume, generate at the largest size the model allows (ChatGPT Image 2.5 takes custom sizes up to a 3840-pixel edge, which covers 12.8 inches at 300 DPI), then enlarge it with Image Upscale 1.0 by a factor of up to 4 when you need a bigger print.

Size rules come from Sume's Image API docs; Image Upscale 1.0 is specified in the OpenAPI document behind the Sume API reference, read on 2026-09-27. The print arithmetic is plain division and applies to any image.

How big can I print an AI image?

Divide each edge in pixels by the DPI your printer asks for. An 8 × 10 inch print at 300 DPI needs 2400 × 3000 pixels. If your print shop accepts 150 DPI, the same file covers twice the width and height.

These are the largest frames ChatGPT Image 2.5 accepts in common print shapes, computed from its published size rules, and what each covers at 300 DPI:

How do I generate an image at print size?

Send the pixels as a custom image_size to openai/gpt-image-2.5. A custom size needs both edges in multiples of 16, a longest edge of at most 3840, an aspect ratio no wider than 3:1, and 655,360 to 8,294,400 pixels in total. The request below asks for a 2560 × 3200 portrait, enough for an 8 × 10 inch print.

• High quality is one of the slow configurations most likely to return 202 with a job to poll instead of the image.
• ChatGPT Image 2.5 estimates each request's price from its size and quality rather than charging one flat rate. usage.cost in the response is the USD amount billed.
• Nano Banana 2 and Nano Banana Pro offer a 4K resolution tier instead of custom sizes. The docs name the tier, not its pixel size, so divide the file's own width and height by 300. 4K AI image generation covers that route.
• Results come back as Sume-hosted, signed URLs, so save the file you plan to print.

curl -X POST "https://api.sume.com/v1/images" \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-image-2.5",
    "prompt": "A botanical poster of wild herbs on cream paper, fine ink lines and soft watercolor washes",
    "image_size": { "width": 2560, "height": 3200 },
    "quality": "high"
  }'

How do I upscale an AI image for print?

Send the image to POST /v1/image-upscale-1.0/upscale with a public HTTPS image_url and an upscale_factor from 1 to 4 (default 2). At a factor of 2, the 2560 × 3200 poster should come back at 5120 × 6400 pixels, about 17.1 × 21.3 inches at 300 DPI.

• output_format is png by default; ask for jpg if your printer wants JPEG files.
• The upscale runs as a job, async by default. Poll it, then read the result artifact: its url, plus width and height when reported, so you can confirm the pixel size before you order a print.
• API pricing lists Image upscale at $0.20 per image, plus a 5.5% agent fee by default, with no separate rates by factor. AI image upscaler API covers the other fields.

curl -X POST https://api.sume.com/v1/image-upscale-1.0/upscale \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "image_url": "https://example.com/botanical-poster.png",
    "upscale_factor": 2,
    "output_format": "png"
  }'

Do I need to set 300 DPI in the file?

Only as a label. DPI is not stored in the pixels: it is a number in the file, or a setting in the print dialog, that tells software how many pixels to spread over each inch. Set it to 300 when you export or print. The pixel count decides how large the print can be at that density, and changing the DPI label alone adds no detail.