Which Sume image models list a 4K resolution tier

Only Nano Banana 2 and Nano Banana Pro list a 4K tier on Sume; Imagen 4 Ultra and Ideogram 4.5 stop at 2K. What 4K costs and when it returns 202.

4 min readSume
All posts

The Sume resolution field accepts normalized tiers: 512, 1K, 2K and 4K. A model accepts only the tiers its catalog descriptor lists, so asking for 4K from a model without one fails with a 400 rather than quietly rendering at 1K.

From the catalog data in the repo, only google/nano-banana-2 and google/nano-banana-pro list 4K. Imagen 4 Ultra and Ideogram 4.5 list 1K and 2K. Higgsfield Soul uses 720p and 1080p instead. ChatGPT Image 2.5 has no tier list: it takes custom pixels on image_size, up to 3840 on an edge.

Tiers and 4K price

Resolution tiers on Sume (read 2026-10-05)
Sume idTiers listed4K billed per image, USD
google/nano-banana-2512, 1K, 2K, 4K0.20
google/nano-banana-pro512, 1K, 2K, 4K0.375
google/imagen-4-ultra1K, 2Kn/a
ideogram/ideogram-v4.51K, 2Kn/a
openai/gpt-image-2.5image_size custom to 3840by tokens

Request and timing

Google's pricing page shows Nano Banana 2 4K at $0.151 per image direct; Sume's list basis is $0.16 for the same tier and $0.20 billed. Nano Banana Pro's 4K list price in the pricing code is $0.30, billed $0.375. Google's image guide lists 512px, 1K, 2K and 4K for Nano Banana 2, so the Sume tiers match the vendor's set.

curl -X POST https://api.sume.com/v1/images \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"google/nano-banana-2","prompt":"Wide mountain panorama","resolution":"4K","aspect_ratio":"16:9","mode":"async"}'

Use async for 4K

The Sume docs say 4K, high quality and large n are the likeliest to exceed the 30 second wait and return a 202 job. Sending mode: "async" makes that the plan rather than a surprise.

How this was checked

Vendor facts come from the pages listed in the sources, read on 2026-10-05. Sume facts come from the Image API docs and the catalog code on main on the same date. Catalogs and limits change, so read the descriptors from GET /v1/images/models before you pin a number in production code.

Sources

Related posts

More in Models

All Models posts

Written by Sume