A release built around iteration
OpenAI announced Images 2.6 on November 4, describing faster generation, more precise editing, greater consistency across revisions, animated GIF workflows and SVG export. The company says generation is 50% faster than the previous version; that is an OpenAI-reported comparison, not an independent benchmark by CherTra News. For low-quality API generation with Flare, OpenAI says some images can be produced in as little as six seconds. Actual time depends on the request and settings.
The API exposes two configurations: GPT-Image-2.6 Flare for faster work and GPT-Image-2.6 Sunburst for higher-detail output. The announcement says a new complexity control affects how much planning the model applies, while quality controls visual fidelity. OpenAI recommends low complexity for simpler edits and medium or high for scenes with more constraints. These controls offer a way to trade speed and elaboration; they do not guarantee exact brand or layout compliance.
GIF and SVG change what teams can produce
The release positions GIF creation as a way to add motion to an image workflow, including short loops for campaigns, reactions and storyboards. GIF is a delivery format, not a substitute for checking frame consistency, loop seams, color reduction, accessibility and file size. A team should preview animation at its intended display dimensions and provide a static alternative when motion is not useful or may distract.
OpenAI also says SVG exports are available for designs intended for websites, apps and print. Vector output can scale cleanly, but the exported file still needs inspection: SVG may contain unexpected paths, unsupported filters, excessive complexity or details that do not match the prompt. Markup-guided editing gives API users a way to point to a region for changes. The changelog lists image generation and editing endpoints, but developers should follow the current image-generation guide for supported inputs and request fields rather than infer API behavior from the launch summary.
Availability and API pricing
OpenAI says Images 2.6 is available in ChatGPT and Codex globally across subscription tiers on desktop, mobile and web. The API provides GPT-Image-2.6 Flare and Sunburst. The current API pricing table lists, for both configurations, image input at $8 per million tokens, image output at $30 per million tokens, and cached image input at $2 per million tokens; text input is listed separately at $5 per million tokens, with cached text input at $1.25. These are token rates, not a fixed price per finished image. Dimensions, output settings and usage determine the resulting bill, so use the official calculator for an estimate and re-check rates before launch.
ChatGPT subscription availability and API billing are separate. An API project needs its own billing and credentials, and an application should handle timeouts, retries and usage limits. Avoid copying model names, prices or request parameters into code as permanent assumptions; version them in configuration and verify the current API documentation.
Where it may help—and what still needs a person
E-commerce teams could explore campaign variants, designers could turn a sketch into a first draft, and developers could request a scalable illustration. Those are candidate workflows, not guaranteed outcomes. Product labels, prices, legal notices, interface states and logos should be checked or placed with deterministic tools. A polished image can still contain inaccurate detail or violate a brand guide.
For a controlled pilot, define which assets may be generated, who approves them, how source files and prompts are retained, and what rights review is required before publication. Compare the full cost of generation, editing and human review with the current workflow. The useful change is a more iterative set of output options; the quality bar, factual checks and permission to publish remain the team’s responsibility.
Sources & further reading
Have a factual correction or a source to suggest? Contact the editorial desk.
