GPT Image 2.5 quality: OpenAI defaults to auto, Sume to high

OpenAI's image guide says both gpt-image-2.5-sunburst and gpt-image-2.5-flare take quality low, medium, high, xhigh, max and auto, with auto as the default. Earlier models stop at high.

Sume's Image API page says that if you omit quality on ChatGPT Image 2.5, the default is high. So a request with no quality field is not the same request on both sides.

What auto does on Sume

If you do send quality: "auto", Sume reserves the cost of max. The page also says auto size and named size presets without a verified GPT pixel mapping reserve the upper bound of output tokens. Both are cost estimates held before the job runs.

Token math at 1024 by 1024

The Sume docs use fal token rates: $30 per million output image tokens, $8 per million input image tokens and $5 per million input text tokens. At 1024 by 1024, xhigh output is $0.09366 and max is $0.21072, before input tokens and before Sume pricing. Input token counts are estimates, and fal rounds the total up to $0.0001.

Pin it

Choose a quality for each job type and send it explicitly: medium for drafts, high for finals, max only where you can see the difference.

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":"Product on white, soft shadow","quality":"medium","image_size":{"width":1024,"height":1024}}'

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.