GPT Image 2.5 Batch API: not supported, so fan out jobs

OpenAI's Sunburst page lists Batch among the endpoints GPT Image 2.5 does not support; only image generation and image edit are supported. To do many images on Sume, submit one POST /v1/images per image with mode: "webhook" or "async" and collect results as jobs finish.

Vendor facts read 2026-09-30 from OpenAI; Sume facts from the Image API docs.

What does OpenAI say is unsupported?

The Sunburst page states only two endpoints are supported, v1/images/generations and v1/images/edits, and that Chat Completions, Realtime, Batch and Fine-tuning are not supported. The Flare page likewise lists just the two endpoints.

How do I run many images on Sume?

Send mode: "async" or mode: "webhook" with a public HTTPS webhook_url. Sume returns 202 with a job envelope holding status_url and result_url. Poll GET /v1/jobs/{id}/status, read GET /v1/jobs/{id}/result, or take the terminal event on your webhook. n also gives several images per call, up to a per-model ceiling read from the catalog.

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": "Catalog shot of a ceramic mug, plain background",
    "mode": "webhook",
    "webhook_url": "https://example.com/hooks/images",
    "metadata": { "sku": "demo-001" }
  }'

What should I watch for?

Use metadata (stored on the job, not sent to the provider) to tie each job to your own record. Verify webhook signatures per the webhooks docs, and keep polling as a backup. Do not assume a batch discount exists; Sume's docs describe none.

How do I check this myself?

Start with a dozen jobs to confirm your webhook handler and result mapping before you submit hundreds. The linked docs pages and the catalog endpoint show the current values, and this post reflects them as of 2026-09-30.