OpenAI Images 2.6: GIFs and SVGs, but no Batch API
OpenAI Images 2.6 adds Flare and Sunburst models with GIF and SVG output, but no Batch API. Keep bulk catalog jobs on 2.5 Batch at half the output price.
OpenAI Images 2.6 shipped on November 4 with two new API models, gpt-image-2.6-flare and gpt-image-2.6-sunburst, plus animated GIF and SVG output. The features are useful. The model pages carry a line that matters more for anyone running image work at volume: Batch is "Not supported." If your catalog pipeline runs overnight in batches, the upgrade costs double on output.
What actually happened
Per the OpenAI API changelog and the image generation guide:
- Two models. OpenAI's docs describe Flare as built "for fast iteration" and Sunburst "for detailed visual work." OpenAI's launch announcement says Flare can return images in as little as six seconds.
- GIF output. Set
output_format: "gif"on/v1/images/generations.frame_countruns 2–64 (default 9) andframe_duration_msruns 20–2000. The edits endpoint and streaming don't support GIF. - SVG output.
output_format: "svg"withn=1, generations only. - Complexity control. A
complexityfield takeslow,mediumorhigh, defaults tohigh, and carries no separate fee. - Markup editing. Draw on a transparent PNG to guide an edit, for both 2.5 and 2.6 models.
The model pages list only the Images generations and edits endpoints. Responses, Chat Completions and Batch are all "Not supported."
On pricing, both 2.6 models match 2.5: $5 per million text input tokens, $8 per million image input tokens, $30 per million image output tokens. The Batch table has no 2.6 rows. It does list the 2.5 models at $15 per million image output tokens, half the standard rate.
Why OpenAI Images 2.6 matters for your business
Split the work by job, not by version number. Overnight jobs that regenerate 2,000 lifestyle shots, alt-image variants or seasonal backgrounds belong on 2.5 through Batch. Work a person waits on, like a merchandiser iterating on a hero image, can use Flare. The extra cost only applies to the calls that need speed.
GIF and SVG remove small, real chores. A short product loop for an email header, or a vector badge for a promo tile, used to mean a second tool or a designer. Now it is one parameter. Check the frame limits before you promise animated PDP media.
There is a deadline in the same docs. The deprecations page shows gpt-image-1-mini, gpt-image-1.5 and chatgpt-image-latest shut down on December 1, 2026. OpenAI's named replacement is 2.5, not 2.6. If anything still calls those IDs, move it now, and to the version that keeps Batch.
Pin the model ID. Each 2.6 model has one snapshot today. Write the exact ID into config so the next release is a decision, not a surprise on your invoice.
Key takeaways
- Images 2.6 adds gpt-image-2.6-flare and gpt-image-2.6-sunburst with GIF, SVG, complexity and markup editing
- Neither 2.6 model supports the Batch API, Responses or Chat Completions
- Standard prices match 2.5, but 2.5 Batch output is $15 per million tokens against $30 for 2.6
- gpt-image-1-mini, gpt-image-1.5 and chatgpt-image-latest shut down December 1, 2026; OpenAI points to 2.5
- Route bulk jobs to 2.5 Batch and interactive work to Flare
Know what each image call costs before you scale it. We build catalog pipelines that route each job to the right model and price tier. Run your volume through our ROI calculator.
Sources: OpenAI API changelog, OpenAI pricing, OpenAI image generation guide, OpenAI deprecations.
- #openai
- #gpt-image-2-6
- #image-generation
- #api-pricing
- #product-imagery
Tommy Rush — Founder, Rush Commerce
Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More
Get The Rush Report weekly — one email, zero fluff.
Keep reading
Shopify Overexposed: scan the live cloud, not your repo
Shopify's Overexposed scanner finds GKE and Cloud Run endpoints open to the internet by probing the live cloud. How to copy the method on a small budget.
Read itHex 4.0 and $100M: BI rebuilt for AI agents, as code
Hex raised $100M at a $1.2B valuation and rewrote its analytics platform for AI agents, with everything as plain code. Why your metrics should live in code you own.
Read it