OpenAI logo symbol used for a FreeAnswers power-user guide to ChatGPT Images 2.6

ChatGPT Images 2.6: Power-User Guide to GIF Creation, API Models and Pricing

by

Last updated: November 5, 2026

ChatGPT Images 2.6, announced by OpenAI on November 4, 2026, speeds up image generation, preserves details better across edits, and adds GIF creation plus saved reference images in ChatGPT. On the API, gpt-image-2.6-sunburst and gpt-image-2.6-flare add complexity control, markup-guided edits, SVG output, and animated GIF output.

Contents

OpenAI published ChatGPT Images 2.6 in the ChatGPT release notes and listed gpt-image-2.6-sunburst and gpt-image-2.6-flare in the API changelog on November 4, 2026. This guide covers both surfaces. Related OpenAI coverage: ChatGPT Slides and ChatGPT API credits.

What’s new at a glance

Item Detail Where
ChatGPT model Faster generation; better detail preservation across edits ChatGPT Images
GIF creation Looping animations; adjust speed; share GIFs Rolling out in ChatGPT
Saved references Save a photo; reuse with @Name in later chats ChatGPT Images
API models gpt-image-2.6-sunburst (detailed); gpt-image-2.6-flare (fast) Generations and edits
complexity low, medium, high (default high); no separate fee Image API
New outputs SVG and animated GIF via output_format POST /v1/images/generations
Markup edits Transparent annotation overlay as markup POST /v1/images/edits

Using Images 2.6 in ChatGPT

The November 4 ChatGPT release notes say Images 2.6 creates images faster and better preserves details across edits, such as keeping a character’s outfit when you change the background.

GIF creation is rolling out. Ask for a looping animation, or upload a sketch and ask ChatGPT to make the character wave. You can adjust playback speed and share the GIF when the feature reaches your account.

Saved reference images let you reuse a photo later. Save a photo as Pepper, then ask for a watercolor of @Pepper without uploading again.

OpenAI did not publish a separate Images 2.6 plan matrix in that note. Image generation still follows your ChatGPT plan’s existing image limits. See ChatGPT usage limits.

API models: Sunburst and Flare

The November 4 API changelog released GPT-Image-2.6 Sunburst and Flare with adjustable complexity, SVG generation, animated GIF generation, and markup-guided editing for supported GPT Image 2.5 and 2.6 models.

  • gpt-image-2.6-sunburst — detailed compositions and editing precision
  • gpt-image-2.6-flare — fast iteration

Both accept text and image inputs. Model pages list support for v1/images/generations and v1/images/edits. Use the Image API directly, or a mainline model with the image generation tool and set the tool model to a 2.6 ID.

The image generation guide says both 2.6 models support complexity, markup edits, SVG, and GIF through the Image API. complexity values are low, medium, and high (default high). Lower values suit focused edits; higher values help dense text and complex layouts. Complexity has no separate fee. quality remains a separate rendering setting.

How to generate a GIF or SVG with the Image API

  1. Create an API key and complete organization verification if the console requires it for GPT Image models.
  2. Call POST /v1/images/generations with model set to gpt-image-2.6-sunburst or gpt-image-2.6-flare.
  3. For SVG, set output_format to svg and n to 1. Decode data[0].b64_json as UTF-8 XML and save a .svg file. SVG does not support streaming, partial images, output_compression, or prompt_rewriter; edits do not return SVG.
  4. For an animated GIF, set output_format to gif. Optionally set frame_count (2–64, default 9) and frame_duration_ms (20–2000, default 200). Describe subject, motion, and loop in the prompt. GIF is generation-only.
  5. Tune quality (low, medium, high, xhigh, max, or auto), size, background, and complexity. Include a privacy-preserving safety_identifier when serving end users.
  6. Check usage tokens against the published text and image rates before scaling traffic.

Prompting patterns are in the image prompting guide. See also our GPT-6.1 Terra guide.

Markup-guided edits and parameters

For markup edits on POST /v1/images/edits, keep the original as the first image, add a same-size transparent PNG of circles or sketches as markup, and describe the change in prompt. Do not combine markup with mask or prompt_rewriter. Each of the first image and annotation layer can contain at most 16,777,216 pixels; the annotation file has a 50 MB limit.

Recommended sizes include 1024×1024, 1536×1024, and 1024×1536. Custom WIDTHxHEIGHT values must be multiples of 16 within documented aspect and pixel budgets. Use background: "transparent" with png or webp. Streaming and partial images are available for supported raster formats, not SVG or GIF.

Pricing and plan notes

Verified on both model pages on November 5, 2026:

Token type Input Cached input Output
Text $5 / 1M $1.25 / 1M —
Image $8 / 1M $2 / 1M $30 / 1M

Rate limits vary by usage tier. Eligible ChatGPT Pro and Business subscribers can claim promotional API credits; see API credits and OpenAI API pricing. ChatGPT GIF and reference features follow ChatGPT rollout and plan image limits, not these API token tables.

Images 2.6 vs Images 2.5

Images 2.5 (September 8, 2026) added sharper details, templates, sketch-to-image, and xhigh/max quality on the API. Images 2.6 keeps that quality ladder and adds ChatGPT GIF creation, saved @Name references, API complexity, markup-guided edits, and native SVG plus animated GIF generation. Prefer Flare for latency and Sunburst for edit fidelity; lower complexity when high planning adds unwanted changes.

Limits and what to verify

  • GIF creation in ChatGPT is rolling out; confirm it appears before relying on it.
  • SVG and GIF are generation-only; edit endpoints do not return those formats.
  • Complex prompts can take up to about two minutes; precise text and recurring characters can still fail.
  • Do not send markup with mask or prompt_rewriter.
  • Organization verification may be required before GPT Image models work.
  • Re-check live model pages for token prices and rate limits.

Frequently asked questions

What is ChatGPT Images 2.6?

ChatGPT Images 2.6 is OpenAI’s November 4, 2026 image release. In ChatGPT it generates images faster and preserves details better across edits. On the API it ships as gpt-image-2.6-sunburst for detailed work and gpt-image-2.6-flare for fast iteration.

How do I create a GIF in ChatGPT Images 2.6?

Ask ChatGPT for a looping animation, or upload a sketch and ask it to animate a character. OpenAI says GIF creation is rolling out; you can adjust playback speed and share the finished GIF when the feature reaches your account.

What are gpt-image-2.6-sunburst and gpt-image-2.6-flare?

They are the GPT Image 2.6 API model IDs. Sunburst targets detailed compositions and editing precision. Flare targets fast iteration. Both support complexity control, markup-guided edits, SVG output, and animated GIF output through the Image API.

How much does GPT Image 2.6 cost on the API?

OpenAI’s model pages list the same token rates for Sunburst and Flare: text input $5 per 1M tokens ($1.25 cached), image input $8 per 1M ($2 cached), and image output $30 per 1M tokens. Complexity has no separate fee.

What is the complexity parameter in GPT Image 2.6?

complexity controls how much the model plans before rendering. Supported values are low, medium, and high; the default is high when you omit it. Use lower complexity for focused edits and higher complexity for dense text or complex layouts. quality remains a separate rendering setting.

How do saved reference images work in ChatGPT?

Save a photo once, give it a name such as Pepper, then reuse it later with @Pepper without uploading again. OpenAI’s ChatGPT release notes introduced this with Images 2.6 on November 4, 2026.

Sources

Image: OpenAI logo, public domain, via Wikimedia Commons (File:OpenAI logo 2025 (symbol).svg). OpenAI and ChatGPT are trademarks of OpenAI; FreeAnswers is not affiliated with OpenAI.



Related answers

Leave a Comment