Create images with GPT Image 2.6, Nano Banana 2.1, and Seedream 5.5
Capriole AI offers free GPT Image 2.6, Nano Banana 2.1, and Seedream 5.5 generation in Browser Chat for signed-in Free, Premium, and Team accounts using Server Storage. Choose any of them in Image mode, or ask the Text Agent while you chat. The image generation itself does not consume your API token allowance; Text Agent messages still follow normal chat limits.- Sign in and use Server Storage: in Account → Settings, leave Enable local storage and Enable No Storage off.
- Open the model selector, choose the Image tab, and select GPT Image 2.6, Nano Banana 2.1, or Seedream 5.5 (GPT Image 2.5 is still available). Describe the image you want.
- Choose an aspect ratio, Resolution, and a batch of 1–4 images, then send. The ratios are Auto (the default) and 1:1, 21:9, 16:9, 3:2, 4:3, 9:16, 2:3, and 3:4; resolutions are 1K, 2K, and 4K, or 1K and 2K with Seedream 5.5. With Auto, an edit keeps the shape of the image being edited. For a new image, GPT Image picks a shape that suits the prompt at 1K and makes a square at 2K and 4K, and Nano Banana 2.1 makes a square.
- Wait for completion in the thread. You can reopen the saved conversation, view each completed image, inspect Info, or download it. Under an answer that is just one image, Copy image copies it at once. When the answer also has text or has several images, Copy opens a menu with Copy text (when there is text), each image, and Copy all, which copies everything together.
@image1, @image2, and so on. When you hover over a generated image, an Edit button appears that puts it into the composer at once so you can describe the next change. On touch screens, the image’s … menu has Edit image and Download image. GPT Image 2.6, GPT Image 2.5, Nano Banana 2.1, and Seedream 5.5 all edit, and Capriole AI uses the model you selected.
You can have up to four active images per account. Every image, edited or new, counts toward that limit. In Text mode, ask the Text Agent to create an image, or to change an attached image or one generated earlier in the chat, instead of switching tabs; it uses the same image-generation service. It uses the image model you name, otherwise the model of the chat’s latest image, and otherwise the default image model, GPT Image 2.6, the first in the Image tab.
For each model’s official capabilities, see OpenAI’s model documentation, Google’s model documentation, and ByteDance Seed’s Seedream 5.5 page. The features available in Capriole AI are described above.
Check the file before uploading
You can attach up to 10 files to one message. Large combined context can still exceed the selected model’s input limits.
Use the chat storage guide before changing a conversation’s storage boundary.
Ask a model to work from a file
1
Open a synced chat
Sign in to Capriole AI. Leave Enable local storage and Enable No Storage off so the chat uses default account-synced storage.
2
Add the file
Choose Add photos and files in the composer. You can also drag files into the chat area or paste a file from the clipboard.Wait for the attachment to finish uploading before sending the message. Capriole shows a processing status while it prepares uploaded files for the selected model.
3
Describe the job, not only the file
Tell the model what to inspect and what output would be useful. For example:For a document review, name the section, decision, or inconsistency you care about. A bare request such as
summarize this gives the model less direction.4
Check the answer against the attachment
Confirm important numbers, quotations, and row references in the original file. Different models have different file and context limits, and converted documents can lose layout details that depend on visual positioning.

