Discover OpenAI’s GPT Image 2.6, its speed and editing improvements, Flare vs Sunburst, and access to both models on CometAPI at 20% lower token prices.
Deon GoodwinAI model and API research team
Updated Nov 5, 2026 10 min read
Discover OpenAI’s GPT Image 2.6, its speed and editing improvements, Flare vs Sunburst, and access to both models on CometAPI at 20% lower token prices.
Use this summary for a fast answer, then read the guide for context, examples and implementation details.
GPT Image 2.6 is OpenAI’s image-generation and editing model family, released on November 4, 2026. It is available through two API models: Flare for fast creative iteration and Sunburst for work that demands greater detail and editing precision. OpenAI’s launch announcement confirms the release, superseding the earlier reports based on leaked API references.
Both GPT Image 2.6 variants are now available through CometAPI at 20% below the corresponding official token rates. The GPT Image 2.6 model brings Sunburst and Flare together and API guide, making it straightforward to choose between detailed visual work and faster creative iteration.
Key Takeaways
The update brings faster generation and more control when refining existing visuals.
Reviewed for clarity, source attribution and current API terminology.
OpenAI has released Images 2.6.
Flare and Sunburst serve different creative priorities. Choose Flare for rapid exploration and Sunburst for demanding detail and editing precision.
The advances go beyond a single finished picture. Visual markup, SVG export, and animated GIFs expand how an idea can be developed and delivered.
Both variants are available on CometAPI. Use gpt-image-2.6-flare or gpt-image-2.6-sunburst.
Choose the variant around the work. Flare suits interactive exploration; Sunburst is positioned for more demanding visual detail and edits. Markup adds visual instructions to an edit. Keep the source image, transparent overlay, and written prompt separate so each can be reviewed.
Compare the finished asset, not just generation speed. Count retries, review time, and preservation of details that matter to your brief.
CometAPI supports GPT Image 2.6 . Use its published model ID to get started, and distinguish CometAPI billing from OpenAI’s direct API rates. CometAPI pricing is 20% below the corresponding official token rates. Actual image costs depend on usage and the settings you choose.
What Is GPT Image 2.6?
GPT Image 2.6 creates and edits images from text and image inputs. The model family includes gpt-image-2.6-flare and gpt-image-2.6-sunburst. A written brief can describe a new composition, while an existing photograph or design can become the starting point for an edit.
The distinction matters when you already have a useful asset. Starting again from a text description can mean rebuilding the subject, framing, and visual style. An editing workflow instead starts with what should remain and specifies what needs to change. For a product team, that might mean revising a label without losing the recognizable package; for a designer, it might mean exploring a layout while keeping the same visual direction.
GPT Image 2.6 Flare vs Sunburst
For a social post or an early campaign concept, fast feedback can help you explore more directions before choosing one. For a product hero image or an edit that must preserve recognizable details, give Sunburst a place in the comparison. The useful choice depends on the asset you need to deliver, rather than a single model being the best fit for every stage.
From the October Leak to the Official Release
An early public reference to GPT Image 2.6 appeared on October 28, 2026, when an OpenAI OpenAPI change added markup to the image-editing interface. At that point, it was an unreleased model associated with native support for transparent PNG annotation overlays. The later November release establishes its current product status.
The Original API Screenshot Shared on X
On October 30, 2026, lifcc (@mylifcc) shared the leak on X, identifying OpenAI’s official API documentation as the source. The post’s accompanying screenshot shows the markup entry on the “Edit an Image” page, with 2.6 highlighted in the supported-model wording. It provides a visual record of the reference described in the repository history.
What Are GPT Image 2.6’s Main upgrade?
The release centers on a familiar creative challenge: getting from an idea to a usable visual without spending every revision rebuilding what already works. Faster drafts help you choose a direction, while more precise edits help you carry that direction through to a finished asset.
More Consistent Characters, Products, and Brand Details
A useful image editor needs to preserve the details that make an asset recognizable while changing the parts you specify. That is especially relevant to a campaign built around one product or character: a new background should still leave the audience looking at the same subject. GPT Image 2.6’s emphasis on editing consistency makes this a central part of the release.
For readers, the benefit is easy to picture. You might explore a seasonal setting for a product photograph, revise a poster’s composition, or adapt a visual for another placement. Compare the subject, lettering, and brand details before approving the result; visual consistency is valuable precisely because those details carry across the campaign.
Visual Markup Makes Local Changes Easier to Explain
Markup lets you point to the part of an image you want changed and explain the change in words. A circle can identify a vase to recolor; an arrow can clarify a proposed position. This makes editing more approachable when a location is easier to show than to describe. The official example below illustrates the idea without requiring a detailed API walkthrough.
More Control Over Simple Edits and Complex Scenes
Quality and complexity serve different purposes: quality concerns visual detail, while complexity controls the planning behind a scene. A small adjustment and a composition with many constraints do not need the same approach. The practical benefit is being able to match the effort to the visual task, rather than treating every request as equally demanding.
SVG and GIF Broaden the Deliverables
SVG exports and animated GIFs expand the kinds of assets you can create with GPT Image 2.6. SVG suits scalable graphics, while GIF adds motion to a short visual idea. These options make the release relevant to more than still-image generation.
For a website or presentation, a scalable illustration can be useful across different placements. For social media, a short loop can bring a mascot or reaction to life. The choice starts with where the asset will appear and how people will view it. A strong static visual and a readable animation solve different communication problems.
How Much Faster Is GPT Image 2.6?
OpenAI reports 50% faster image generation than before. Flare can generate an image in as little as six seconds at low quality, while Sunburst emphasizes detail and consistency across edits at medium or high quality. These are the conditions described in OpenAI’s release announcement; individual response times vary.
Faster generation matters most when you are still choosing a direction. Shorter waits can make it easier to compare a composition, revise a brief, and try another approach. For a finished asset, editing precision may matter more than the fastest first draft. Think of the two variants as choices for different creative priorities, then judge the result against the image you need.
What Does GPT Image 2.6 Cost and why should choose CometAPI?
CometAPI offers both GPT Image 2.6 variants at 20% below the corresponding official token rates. Its updated model and pricing page lists the following prices for both gpt-image-2.6-sunburst and gpt-image-2.6-flare. All figures are US dollars per million tokens, checked on November 5, 2026.
These are usage-based token prices, not fixed prices per image. The final cost depends on the inputs, output settings, and any additional revisions. Cached rates apply to eligible cached usage. For recurring creative work, the most useful budget is the cost of the images you actually approve and use.
Where Can GPT Image 2.6 Help Creators?
The strongest use cases start with a specific deliverable: a product image, a campaign visual, a design concept, or an animated asset. Decide what the audience needs to see, then use generation and editing to develop that idea. The examples below are practical ways to explore the model family.
Product Photography
A product team can explore different settings and compositions around an existing product photograph. Keep the package, label, and recognizable shape at the center of the brief. Flare can help compare directions quickly; Sunburst is worth testing when the final visual depends on more delicate details.
UI and Graphic Design
Designers can explore a visual direction before committing to the finished layout. A generated mockup can make an idea easier to discuss, while an SVG may provide a scalable asset for a later stage. Check the result in its intended placement: readable text and a clear hierarchy matter more than an attractive thumbnail.
Marketing Asset Production
Marketing teams frequently need multiple versions of the same visual asset. A useful test is to prepare several campaign briefs against one reference image, then compare whether the product identity and brand treatment remain consistent across the results.
For example, one campaign might prioritize a headline, another might leave space for a price badge, and a third might need a different composition. Judge each variation against its placement and copy requirements, then record how many revisions were needed to make it usable.
Where Can You Use GPT Image 2.6?
OpenAI announced Images 2.6 for ChatGPT and Codex across subscription plans, alongside the Flare and Sunburst API models. For application development, start with the exact model identifier rather than a generic family name, and check the account’s current access and limits before planning a rollout.
Direct OpenAI API Access
For a new image, use the image-generation route; for a reference-based change, use the image-editing route. Keep the model, prompt, source assets, and request settings together so a successful experiment can be reproduced. Test one complete request before expanding into batch production or a customer-facing editor. The Sunburst model page and Flare model page provide the authoritative identifiers and model descriptions.
When an integration fails, distinguish a model-access problem from an input problem. Check the response error, then verify the endpoint, parameter combinations, and files supplied. An old SDK or a wrapper that does not pass through a newer parameter can also change what the service receives. Keep the first test simple enough that you can inspect each part of the request.
GPT Image 2.6 Is Available on CometAPI
You can access both variants through CometAPI’s unified GPT Image 2.6 API. The two model identifiers are:
gpt-image-2.6-flare — for faster generation and interactive creative work.
gpt-image-2.6-sunburst — for detailed visuals and precise editing.
Sign in to CometAPI, choose the variant that fits your project, and follow the model page’s access instructions. Both belong to the same model family and share the discounted pricing shown above. You can begin with a real brief, compare the results, and build from the version that best fits your intended use.
Images 2.6 is also available in ChatGPT and Codex, alongside OpenAI’s direct API. For readers already using CometAPI to access image models, the new listing provides another option within that existing workflow.
Conclusion
GPT Image 2.6 brings faster generation, more precise editing, and additional ways to turn a visual idea into a usable asset. Flare is a practical starting point for fast exploration; Sunburst focuses on work where detail and consistency are central to the result.
Both variants are now accessible through CometAPI, with token prices 20% below the corresponding official rates. Start with the kind of visual you actually need, choose the variant around that task, and review the result in its intended setting. The value of the release is in making that creative process faster and more controllable.
Decision
Flare
Sunburst
API model ID
gpt-image-2.6-flare
gpt-image-2.6-sunburst
Official positioning
Interactive generation and editing
Detailed visual work and editing precision
A practical first test
Draft concepts and quick revisions
Demanding compositions and final-asset edits
What to measure
Response time and useful drafts per session
Detail preservation and accepted results
Selection principle
Use when faster feedback helps you choose a direction
Use when the final details justify closer review
Token category
OpenAI official price
CometAPI price
Savings
Text input
$5.00
$4.00
20%
Cached text input
$1.25
$1.00
20%
Image input
$8.00
$6.40
20%
Cached image input
$2.00
$1.60
20%
Image output
$30.00
$24.00
20%
What Is GPT Image 2.6? Features & Access - CometAPI
