OpenAI shipped gpt-image-2 a few days ago, and it is already live in Shortodella. You can pick it from the model selector inside the editor — the option now reads "GPT Image 2".
It replaces gpt-image-1 entirely. If you used the previous GPT Image option, you are already on the new model.
When to Use It
Highest fidelity. gpt-image-2 produces sharper detail and cleaner text rendering than its predecessor.
Better reference handling. Edits stay closer to the source image, so iterative work drifts less.
For everyday work, Nano Banana 2 is still the default. It is faster and cheaper. Switch to GPT Image 2 when you need the extra polish.
Pricing and Sizes
Generation costs 4 energy per image. Available sizes:
• 1024×1024 (square)
• 1024×1536 (portrait)
• 1536×1024 (landscape)
All generations run at high quality. Larger 2K and 4K outputs are not exposed — they cost too much per image and are not yet stable on OpenAI's side.
How to Use
Open the editor, hit the model selector next to the chat input, and pick GPT Image 2. That is it. The agent will route image requests to the new model.
Try it now at shortodella.com.