Image model sume/auto or a pinned id: when to send which

Send sume/auto when you want Sume to choose and pin an id when you need the same look and price every time. Auto never names the model that ran.

5 min readSume
All posts

Pin an id when the look, the price or a parameter matters, and use sume/auto when you do not care which model runs. sume/auto is a Sume-only model value: Sume selects the family and never discloses which one ran. GET /v1/images/models does not list it, and the response and job.model echo sume/auto, so you cannot A/B test models with it.

Differences in practice

The table is from the public Image API docs.

Sume Image API docs and repo, read 2026-10-13
Questionsume/autoPinned catalog id
Which model ranNot disclosedThe id you sent
Listed in /v1/images/modelsNoYes
Parameter checkOnly what the routed model acceptsPer-model descriptors; 400 if unlisted
PriceEstimate on the rate cardRead from the model's endpoints record
MCP defaultYes, payload.model is optionalSet payload.model
REST defaultNone; model is requiredNone; model is required

A safe pattern

Start with auto while exploring prompts, because you do not have to pick an id. When a result is good, ask for the same output again with an explicit id only if you can tell which one matches, which you cannot from the response. So for repeatable assets, pin from the start and keep the id in config.

Log the model you sent and the model the job stored. Retired ids, such as google/nano-banana-2, run as their successor and the job stores the successor id.

curl -s https://api.sume.com/v1/images \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"sume/auto","prompt":"minimal logo mark, teal"}'

Remember

REST requires model; omitting it is an error. Only the MCP tool fills in sume/auto for you. See the rate card for the sume/image-router estimate.

Sources

Related posts

More in Developers

All Developers posts

Written by Sume