GPT-Image 2.6
Generate and edit images with GPT Image 2.6 on Mitte. About twice as fast as 2.5, with animated GIFs, SVG vectors, markup edits, real text and references that stay recognizable.
GPT Image 2.6 is OpenAI’s newest image model, and you can use it on Mitte right now. Describe an image and get it. Attach a photo and say what to change. It does everything GPT Image 2.5 did, about twice as fast, and it opens three doors that were closed before: animated GIFs, editable SVG vectors, and edits you direct by drawing on the picture.
What made the earlier model good is still here and sharper. Put a person from a reference photo into a new scene and they still look like themselves. Swap a product label and the hands holding it do not change. Ask for a poster with specific copy and the words come out spelled right, in the layout you described. The Standard tier returns results in as little as six seconds, so you can iterate the way you would sketch. When a piece needs every detail right, switch to the Slow tier for finer rendering and edits that stay consistent from one pass to the next.
What can you do with it?
- Generate from a brief. Describe the subject, style, lighting and composition, up to 4 images per run. Long prompts are welcome; write it like a creative brief, not a search query.
- Edit with plain language. Attach an image and type “make it golden hour”, “remove the second chair”, “change the jacket to red”. Everything you did not mention stays put.
- Point at the change. Draw a circle around the lamp and write “replace with a floor plant” on a transparent overlay. GPT Image 2.6 reads your markup and edits exactly there, leaving the rest of the room alone.
- Make an animated GIF. Turn a prompt or a still image into a short loop: a paper plane looping once, a logo that breathes, a product turning on its stand. Set the frame count and the speed.
- Export real vectors. Generate a logo, icon or diagram as an SVG with editable paths, shapes and colors. Drop it into a website, an app or a print layout and scale it to any size.
- Keep people and products recognizable. Use reference photos of a face, a product or a place, and move them into new settings, seasons or art styles.
- Combine up to 16 references. Product from one shot, model from another, environment from a third. Name each one in the prompt and say what it contributes.
- Mask an exact area. Add a mask to lock the edit to one region: a headline, a sleeve, a window.
- Design with real text. Posters, packaging mockups, menus, social tiles, thumbnails. Put the copy in quotes and it renders it.
- Cut out on transparent. Generate logos, stickers and product shots on a transparent background, ready to drop into a layout.
How to use it
- Write the prompt. Subject, style, lens, light and any copy in quotes. The more specific, the less it drifts.
- Attach references to edit (optional). One image for a straight edit, several when you are combining elements. Your prompt becomes the instruction for what changes.
- Add a mask or a markup (optional). With a reference attached, upload a black-and-white mask where white marks the area to change, or a transparent PNG with your circles, arrows and notes drawn over the image. Use one or the other.
- Choose ratio and resolution. Square, landscape, portrait or cinematic 21:9, at 1K, 2K or 4K. Editing? Pick Auto and the output matches your source image.
- Pick a quality level. Low for quick explorations, High for finished work, Extra High or Max when every detail counts.
- Pick a speed. Standard for most work. Slow when fine detail, small type or a precise edit has to be perfect and you can wait.
- Pick an output. JPEG, PNG or WebP for stills. GIF for an animation, with frame count and frame duration in the settings. SVG for a vector logo or icon.
- Generate. Download the results, share them, or keep editing right on Mitte.
What you get
Up to 4 images per run in JPEG, PNG or WebP, at 1K, 2K or 4K, in your chosen ratio. Transparent backgrounds come through in PNG and WebP. GIF output gives you an animation of 2 to 16 frames. SVG output gives you one scalable vector file with editable paths.
Tips
- Explore fast, finish slow. Run a batch of 4 on Standard at Low quality to find the direction, then rerun the winner on Slow at High or Extra High with the same prompt.
- Say what stays. “Replace the label with the attached artwork. Keep the lighting, hands and background.” Naming what should not move is the fastest way to a clean edit.
- Draw instead of describing. When the change is about a place in the image, a circle and an arrow on a markup overlay beat three sentences of spatial description.
- Turn planning down for small edits. The Planning setting defaults to High. For a focused change like swapping a sign or a background, set it to Low so the model changes less and finishes faster. Keep High for dense text and busy layouts.
- One change per pass. Three small edits in a row beat one paragraph of simultaneous changes.
- Address references by number. “The person in image 1, wearing the jacket from image 2, on the street in image 3.”
- Quote your copy. Put headlines and labels in quotation marks and tell it the font feel: “bold condensed sans”, “hand-lettered script”.
- Describe a GIF as one loop. “A coffee cup steaming, steam rising and fading, seamless loop.” Nine frames at 200 milliseconds is a good start; fewer frames read as a flipbook, more as motion.
- Keep SVGs flat. Vectors love simple shapes, solid fills and clean type. Ask for “flat vector logo, three colors, no gradients” and you will get paths you can actually edit.
- Transparent needs PNG or WebP. JPEG cannot hold transparency. Switch the output format before you generate a cutout.
- Save 4K for the final. Lock the composition at 1K or 2K, then rerun the keeper at 4K.