GPT Image 2.5 400 unsupported_parameter: fields each model accepts

A 400 unsupported_parameter from POST /v1/images means you sent a field that the selected model does not list in its catalog. Sume rejects it instead of dropping it silently. For the GPT Image family, the usual causes are mask_url or background sent to openai/gpt-image-2, or output_compression and seed, which no model advertises yet.

Facts are from Sume's Image API docs, read 2026-09-29.

Which GPT Image fields does each model list?

How do I see what a model accepts before I call?

Read its endpoint record. supported_parameters there is the definitive set, and model-level and endpoint-level sets are identical because Sume serves each model through one sume endpoint.

curl "https://api.sume.com/v1/images/models/openai/gpt-image-2.5/endpoints" \
  -H "Authorization: Bearer $SUME_API_KEY"

What about output_format svg or provider options?

The docs list png, jpeg, webp or svg as schema values, but the catalog is the authority for a model: check output_format in its descriptor. provider.options must be omitted or empty, since allowed_passthrough_parameters is empty for every endpoint in v1.