GPT Image 2.6 Sunburst API

What is new

• Released by OpenAI on 2026-11-04. Same token rates as GPT Image 2.5 ($30 per 1M image output tokens); high, xhigh and max use the same output tokens per image, while low and medium use 4.2× and 3.3× more tokens.
• Exact output sizes: 1K, 2K and 4K at any aspect ratio from 21:9 to 9:21 (4K is 3840×2160 at 16:9).
• markup_image: draw circles, arrows or notes on a transparent layer and the edit follows them.
• complexity (low, medium, high) sets how much the model plans, separately from quality.
• PNG, JPEG, WebP and native SVG vector output (output_format: "svg", n: 1).

Pricing

The prices above are for a request without reference images, videos or audio. When the provider reports what it charged for the call, you pay that amount times the same markup as the listed price, so reference inputs, longer inputs and extra outputs cost more and unused options cost less. We hold the highest possible amount when the task starts (reference videos at their maximum length) and refund the rest when it finishes. Calls served by fal add $0.019 per reference image.

Compared with fal

Options where KeelRouter costs less for GPT Image 2.6 Sunburst.

fal list prices from its model pages, checked 2026-11-05; fal prices some sizes differently.

Async: POST /v1/images/generations returns a taskId; poll the same path with ?task_id= or pass a callback_url.

curl https://api.keelrouter.com/v1/images/generations \
  -H "Authorization: Bearer $KEELROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "gpt-image-2.6-sunburst",
  "prompt": "A lighthouse at dawn, watercolor",
  "resolution": "1K",
  "quality": "low"
}'

Parameters

FAQ

How much does the GPT Image 2.6 Sunburst API cost on KeelRouter?

From $0.025 per image. You pay per call from a prepaid USD balance; failed calls are not charged. The prices above are for a request without reference images, videos or audio. When the provider reports what it charged for the call, you pay that amount times the same markup as the listed price, so reference inputs, longer inputs and extra outputs cost more and unused options cost less. We hold the highest possible amount when the task starts (reference videos at their maximum length) and refund the rest when it finishes. Calls served by fal add $0.019 per reference image.

Is GPT Image 2.6 Sunburst cheaper on KeelRouter than on fal?

For these options, yes: 1K max $0.21 vs $0.21072 per image; 2K xhigh $0.19 vs $0.19029 per image. fal list prices from its model pages, checked 2026-11-05; fal prices some sizes differently. No subscription; top up from $5 and pay per call.

How do I call GPT Image 2.6 Sunburst?

POST to /v1/images/generations with model "gpt-image-2.6-sunburst". You get a taskId back; poll the same endpoint with ?task_id= or pass a callback_url.

Is this the official GPT Image 2.6 Sunburst API?

No. KeelRouter is an independent API gateway. Requests run on GPT Image 2.6 Sunburst through our upstream providers, and the image model keeps its own safety filters.

How long are results kept?

Result files are kept for 30 days, then deleted. Download what you want to keep.

More from OpenAI

• GPT Image 2

  From $0.011 per image

• GPT Image 2 All

  From $0.0068 per image

• From $0.0078 per image

• GPT Image 2.5 Flare

  From $0.0044 per image

• GPT Image 2.5 Sunburst

  From $0.0068 per image

• GPT Image 2.6 Flare

  From $0.025 per image