Luma uni-1 API: image models, types, and what Sume lists

The Luma Agents API has two image models, uni-1 and uni-1-max, and both support the image and image_edit types at 2048-pixel output. Sume does not use these ids: its image catalog is its own list, so read GET /v1/images/models to see what you can call.

Luma facts are from its Models and Pricing page; Sume facts are from Image models, read 2026-10-01.

What are uni-1 and uni-1-max?

Image output is 2048 px (2K) in png or jpeg.

How do I call uni-1?

Luma sends one POST /v1/generations request with model, prompt and a type, then polls for the result. The same endpoint also serves its ray-3.2 video model, with different supported type values.

Results come back as presigned URLs that the page says expire after 1 hour.

Is uni-1 on Sume?

No Luma id appears in the catalog code I read; it defines ids such as grok-image and flux-2-pro. The live answer is the models endpoint, which also returns pricing. Sume adds one more value, model: "sume/auto", where Sume picks the family and never discloses which one ran.

Each model has one sume endpoint in v1.

curl "https://api.sume.com/v1/images/models" \
  -H "Authorization: Bearer $SUME_API_KEY"

What should I do if I need uni-1 specifically?

Call Luma directly for it and use Sume for the models it lists. Mixing is fine at the application level: keep the prompt and your own job record, and swap the model id and endpoint per call. See image generation with reference images for the Sume side of an edit.