500 images for a a prompt library: GPT Image 2.5 cost by quality

500 images at 1024x1024 cost $12.3750 at quality: low, $27.8125 at medium and $111.1875 at high with ChatGPT Image 2.5 (openai/gpt-image-2.5) on Sume. That is one image per request, priced from Fal's output-token rates with Sume's margin.

Sume's Image API docs (read 2026-10-06) say that if you omit quality the default is high, and that auto reserves the max price. For a a prompt library, pass the quality you mean.

What does a a prompt library cost at each quality?

Per-image prices are computed for a 1024x1024 output with no reference images.

How do I set the quality?

quality accepts auto, low, medium, high, xhigh and max. Draft at low, rerun the keepers at high.

curl -X POST https://api.sume.com/v1/images \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: img-video-prompt-reference-library-001" \
  -d '{
    "model": "openai/gpt-image-2.5",
    "prompt": "Studio photo of a ceramic mug on a pale wood table",
    "quality": "low"
  }'

What changes the price?

Check the live per-endpoint price at GET /v1/images/models/{model_id}/endpoints before a large batch.

• Size: a larger output uses more output tokens; the table is for 1024x1024.
• References: up to 16 input images add estimated input-token cost.
• auto quality: reserves max at submit.