Skip to main content
OpenAIImage model

GPT Image 2.5

OpenAI's newest image model. Photoreal shots, clean text and edits that change only what you ask.

Build with GPT Image 2.5

Each clip starts as one GPT Image 2.5 image, animated with Seedance 2.5 in the same Runable task.

GPT Image 2.5 on Runable

  • Pick Flare or Sunburst

    Both are in the image menu, at the same price.

  • Edit with references

    Add up to 16 images and say what to change.

  • No OpenAI account

    Runs on your Runable credits, next to other image models.

How to use GPT Image 2.5 on Runable

  1. Select GPT Image 2.5

    Start a task on Runable and pick GPT Image 2.5 in the model menu.

  2. Describe the image

    Type what you want to see, or add photos for it to follow.

  3. Turn it into a video

    Refine the image by chatting, then animate it with any video model on Runable.

One prompt, finished work

Real results from single Runable tasks. Pick one to watch it.

Made on Runable with GPT Image 2.5 and Seedance 2.5
Real estate photo of a small Scandinavian kitchen: oak cabinets, white terrazzo counter, a bowl of lemons, morning light through a tall window, wide-angle 24mm.
Sunburst on LMArena text-to-image, Oct 2026
#1
lower latency than GPT Image 2 (Flare)
50%
reference images per edit
16
turn to build
Your

Why choose GPT Image 2.5

  • Edits only what you ask

    OpenAI says it keeps the rest of the image the same.

  • Faces stay recognizable

    Better at keeping people from your reference photos.

  • Holds up over many edits

    Earlier changes stay put as you keep refining.

  • Text that reads

    Posters, menus and UI mockups with the words you wrote.

  • Two speeds

    Flare for quick drafts, Sunburst for the final image.

  • Tops the leaderboards

    Sunburst ranks first on Artificial Analysis and LMArena.

GPT Image 2.5 specs

Maker
OpenAI
Released
September 8, 2026
Versions
Flare (fast) and Sunburst (most precise)
Inputs
Text, or up to 16 reference images to edit
Quality
Low, medium or high
Largest size
2560 x 1440
Aspect ratios
1:1, 3:2, 2:3, 4:3, 3:4, 16:9, 9:16
Output
One image per request
Specs are for GPT Image 2.5 on Runable. OpenAI's API adds xhigh and max quality, sizes up to 4K marked experimental, and transparent backgrounds. Flare and Sunburst cost the same on Runable, and you use your plan's credits instead of an OpenAI key.

The only AI agent you need

Compare plansHelp

Need SSO, data localization, premium support, onboarding assistance? Contact Sales

GPT Image 2.5 FAQs

GPT Image 2.5 is OpenAI's image generation and editing model, released on September 8, 2026. In ChatGPT it is called ChatGPT Images 2.5. Developers get it as two models, GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst.

Sources: OpenAI: Introducing ChatGPT Images 2.5 · OpenAI API: GPT-Image-2.5 Sunburst · OpenAI API: GPT-Image-2.5 Flare · Artificial Analysis: text-to-image leaderboard · LMArena: text-to-image leaderboard · Runable: GPT Image 2.5 is live on Runable

More AI models on Runable

Put GPT Image 2.5 to work

Describe the job. Runable runs it on GPT Image 2.5 and ships the result.