MAI-Image-2.6-Flash API: not on Sume, models to use instead

Sume does not list MAI-Image-2.6 or MAI-Image-2.6-Flash in its image catalog, so a request with that id returns 404 model_not_found. Microsoft offers the models in its own MAI Playground and in Microsoft Foundry (Public Preview). On Sume, call GET /v1/images/models and pick from what it returns.

Microsoft's facts are from its announcement; Sume's from the Image API docs. Both read 2026-10-01.

What did Microsoft announce?

The 2026-09-04 post introduces MAI-Image-2.6 and a 2.6-Flash variant. Microsoft quotes Flash at 2.8x the speed of GPT-Image-2-Medium with 72% greater efficiency. Both support multi-reference editing and web grounding, with output up to 1.5K resolution and dynamic aspect ratios. The post gives no price.

Those speed and efficiency numbers are Microsoft's own; this post has not tested them.

Is there a Sume equivalent?

Sume does not claim one. It serves other vendors' models: openai/gpt-image-2.5, google/nano-banana-2, google/nano-banana-pro, black-forest-labs/flux.2-pro, ideogram/ideogram-v3 and more, plus sume/auto, where Sume picks the family and does not tell you which one ran.

Which ids does each side list?

Model ids named in this post.

How do I try a replacement?

Send one request and read the response. usage.cost is the billed USD for that call.

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-pro",
    "prompt": "Storefront sign reading OPEN LATE, rainy street, wide shot",
    "aspect_ratio": "16:9"
  }'

What should I watch for when switching?

Each model accepts only the values its catalog descriptors list, and an unsupported parameter returns 400 unsupported_parameter. Read supported_parameters for the model you picked before copying options from another vendor's playground.