ChatGPT Images 2.5 is here: Flare and Sunburst in Flixly

The API models introduced with ChatGPT Images 2.5 are now available in Flixly: GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst. Create an image, edit an existing one, or build a sequence in Stop Motion Studio.

Start with Flare in the Image Generator or Sunburst in the Image Generator. Each link selects that model so you can describe the image and review its credit quote. This release starts with medium quality at 1024 x 1024. Each request returns one image.

ChatGPT Images 2.5, Flare, and Sunburst

ChatGPT Images 2.5 is the product name used in OpenAI's announcement. GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst are the API models introduced alongside it. Flixly offers those models for generation and editing.

Choose either model to try it with your own brief. Results and generation times depend on the request; review the image before using it in a finished design. We have not published a Flixly speed or quality comparison between the two variants.

The editing links open Image mode. Add your source image and describe the change before generating.

Choose your output

The Image Generator includes background and output-format controls. Use PNG or WebP when you need a transparent background, or JPEG for an opaque image. Review the selected model, settings, and credit quote before submitting.

Medium quality at 1024 x 1024 is the initial supported combination. The underlying API offers more quality and size options, but this release does not make every combination available in Flixly. The generator shows which settings have a current quote.

Prompt example: a transparent illustration

Open Flare, choose a transparent background with PNG or WebP output, and try this prompt:

A single isometric clay-style illustration of a small greenhouse with terracotta pots and three leafy plants inside. Rounded forms, soft green and cream colors, clean edges, centered composition. Isolate the greenhouse on a transparent background. No text, labels, or extra objects.

Generated example: a product photograph with Flare

This image was generated with GPT-Image-2.5 Flare at medium quality, 1024 x 1024, with an opaque background and PNG output, using this prompt:

A matte blue water bottle on a warm cream surface, soft side lighting, clean composition, room above for a headline, no added text.

Open Flare with the model selected to try the prompt yourself, or try it with Sunburst. Check text, edges, fine details, and the composition in each result.

Keep working with Edit this result

When you have an image you want to refine, select Edit this result. Flixly carries it into Image mode and selects the corresponding editing model when available. You can describe the next change without downloading and uploading the result again.

Prompt example: a precise background edit

After generating your own bottle image, choose Edit this result, or open Sunburst editing and add a source image. Then try:

Change only the cream backdrop and tabletop to pale sage green. Keep the bottle's exact position, silhouette, blue finish, cap shape and color, and size in the frame. Preserve the direction and softness of the lighting and the contact shadow. Add no objects or text.

You can run the same instruction with Flare editing. This is an instruction for a focused edit; inspect the result to see which details were preserved.

Image mode supports up to 16 source images in total for these models. The result you are editing counts toward that limit. Additional references can help communicate a palette, scene, or design direction; describe the role of each image in your prompt.

Review the new settings and credit quote before generating the revision. Each edit is a separate paid request, and the model may change details you asked it to preserve.

Turn an idea into a frame sequence

Open Stop Motion Studio with Flare or with Sunburst. Describe the scene and plan an action for each frame, then review the quote for the remaining sequence.

The studio generates the opening frame and creates later frames as sequential edits. It uses the previous frame and, when different, the opening frame as references. Those references help communicate the scene; they do not guarantee identical characters or details across frames.

Watch the three-frame fox animation

This clip uses the Sunburst fox image as its opening frame, followed by two sequential Sunburst edits. The fox lifts a front paw and puts it down a small step forward.

Download the three-frame fox video. It contains three still frames held for half a second each: 1.5 seconds at 2 fps, with no audio or interpolated motion. Press Play to watch it repeat.

To try the movement yourself, keep the camera and lighting fixed and give each frame one action:

1. Frame 1: The fox stands with all four paws on the ground beside the paper tree.
2. Frame 2: The fox lifts the front paw nearest the camera; its body and the tree stay in place.
3. Frame 3: The fox places that paw a small step forward; keep the same framing and background.

The Stop Motion Studio guide includes the starting prompt and a frame-by-frame walkthrough. Inspect your own edits for changes in the character or scene before exporting.

You can pause after the current frame, review the sequence, and revise actions that have not been generated. Each AI frame is a separate paid image request. The studio assembles the finished frames into a silent MP4 at your chosen playback speed. Read the Stop Motion Studio guide for the full workflow.

Questions

Which settings are available at launch?

Generation and editing start with medium quality at 1024 x 1024. The Image Generator also offers background and PNG, JPEG, and WebP output controls. Transparent output requires PNG or WebP. Stop Motion Studio uses medium-quality square PNG frames with an opaque background; its video canvas is a separate setting.

How much does an image cost?

The generator shows the current credit quote for the selected model and supported settings before submission. Creating an image and editing it are separate requests. Stop Motion Studio shows a quote for the remaining frames before you start or resume generation.

Can I edit an image I already have?

Yes. Open Flare editing or Sunburst editing and add your source image. You can use up to 16 source images in total, including references. Use Edit this result to continue from a generated image.

Does this include the ChatGPT editor?

This release provides the Flare and Sunburst API models inside Flixly's image and stop-motion workflows. It does not add region masks, sketch input, or a persistent revision-history editor.

Create your first image with Flare, try Sunburst, or plan a Flare stop-motion sequence.