Ideogram magic_prompt off and JSON prompts vs Sume passthrough

On Ideogram 4.0, magic_prompt set to off keeps your wording and only converts it into a structured prompt. On Sume you cannot send that switch: allowed_passthrough_parameters is empty for every endpoint in v1, and a parameter the model does not list returns 400 unsupported_parameter.

Ideogram facts are from its Ideogram 4.0 generate page; Sume facts are from Image models, read 2026-10-01.

What does each magic_prompt value do on Ideogram?

Can I send magic_prompt to Sume?

No. The docs state that allowed_passthrough_parameters is empty for every endpoint in v1, so provider.options must be omitted or empty. Separately, parameters a model actually accepts are published as capability descriptors per model, and a request that sets one the model does not list is rejected rather than dropped. A provider-specific switch like magic_prompt is therefore a 400, which at least tells you it did nothing.

curl https://api.sume.com/v1/images \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"ideogram/ideogram-v3","prompt":"exactly this wording"}'

How do I keep the prompt as I wrote it?

The docs describe no Sume prompt-rewrite control for this route, so there is no switch to flip. Read supported_parameters for the model and send only what it lists. For how rewritten prompts surface in a response on another model, see the revised_prompt field. Compare the output against your prompt before you rely on a wording-sensitive result such as embedded text.

Which Ideogram model does Sume list?

The code catalog maps ideogram/ideogram-v3 to ideogram-v3. The page above documents Ideogram 4.0, a separate endpoint, so its magic_prompt values are context for the Ideogram API, not a description of Sume's catalog row.