Spaces
Spaces is a node canvas where you connect text, image, video and assistant steps into one workflow. Every node that works, what it takes in, what it makes and what it costs.
For the complete documentation index, see llms.txt. This page is also available as Markdown.
Spaces is at /studio/spaces, listed as My spaces in the sidebar. It is a canvas of nodes: each node does one job, and you connect them so the output of one feeds the next.
#Before you start
- Signed in, your canvas autosaves to the cloud 5 seconds after your last change, and a save is limited to 20 a minute. The My spaces page lists the spaces from all your devices.
- Signed out, the page shows the spaces saved on this device and says "Sign in to see the spaces you made on your other devices."
- The My spaces page has the tabs All, Shared and Mine, and a box that reads Search boards....
- Running a generation node needs a signed-in account and credits.
#Add and connect nodes
Step 1: Open the Add node menu
It has the tabs All, Basics, Image, Video, Media and Utility. Select a node to place it on the canvas.
Step 2: Connect nodes
Drag from a node's output handle on the right to an input handle on the left of another node. Handles carry a type: text, image or video, and a connection is made only between matching types. A mismatched connection is not made, and no message appears.
Step 3: Run a node
Use Run in the node's toolbar. The right-click menu also has Run from here, Focus (F), Rename, Reset Size, Duplicate (CmdD) and Delete.
On the canvas, CmdZ undoes, CmdD duplicates the selected node and Delete removes it. The corner controls are Zoom in, Zoom out, Fit view (F) and Toggle minimap (M).
#The nodes
| Node | Takes in | Makes | What it costs |
|---|---|---|---|
| Text | Nothing. You type in the box, "Describe your vision here...". | Text, to feed a prompt input. | Free. |
| Image Generator | A text prompt, and optionally one reference image. | One to four images. Controls: model (11 image models, GPT Image 2 and Nano Banana 2 among them), aspect ratio 16:9, 9:16 or 1:1, and quantity 1 to 4. | The model's base price times the quantity, shown on the button as Generate (N credits). |
| Video Generator | A prompt and a start image. | One video. Controls: model (Kling 3.0 by default, Luma Ray, Seedance 2.0, Kling 2.5, Wan 2.1), aspect ratio 16:9, 9:16 or 1:1, length (5, 10 or 15 s where the model offers them) and a sound switch that starts off. Kling 2.5 runs only with a start image. | By model, length and sound, shown on the button as Generate (N credits). |
| Assistant | A text input, and optionally an image to describe. | A text reply, with a Copy text button. | 6, 12 or 30 credits a message, by the same rule as the editor's AI chat. The button label shows a fixed estimate; the charge follows the request. |
| List | Text and images. | Text and images, held as a list: up to 10 text items and 12 media items. Keep Items or Discard after run chooses what stays. | Free in itself. The nodes you connect to it charge as usual. |
| Variations | One image. | One new image for each preset you check, from 15: 3/4 View, Aerial, Back View, Closeup, Ext. Long Shot, Extreme Closeup, Eye Level, High Angle, Long Shot, Low Angle, Med. Closeup, Medium Long, OTS, POV and Profile. | 80 credits per variation. Checking 4 presets costs 320. |
| Image Editor | One image. | An edited image. Tools: Crop Image, Expand Image, Remove background, Change camera angle, Adjust and AI Edit. | Crop and Adjust are free. Expand is 170, Remove background 50, Change camera angle 80 and AI Edit 80, shown on the button as Apply (N credits). |
| Upload, Assets, Stock | Nothing. | Your own image or video. A media node shows "Drop an image or click to upload", with Upload media, Browse library and Replace. | Free. |
The Video Generator also shows an End image input, but a video request carries the start image only. The node's other outputs are Remix and Audio.
#Credits in Spaces
Generation nodes call the same image, video, edit and chat services as the tool pages and spend from your one credit pool. A node is priced from the model's base setting, not from the per-setting quote the Create pages use, so the same model can cost a different amount in a node. For example Nano Banana 2 is 70 credits in an Image Generator node and 147 on Create image at its default 1K setting. With a reference image connected, an Image Generator run can also cost more than its button shows. See What actions cost.
If your balance is short, the node reports "Insufficient credits. Need <N> credits." Failed images are refunded. If the node cannot tell whether a request went through, it says to check your Library and credit balance before retrying.
#Templates
The template gallery is organised by category: All, New, Featured, Branding, Social media, Advertising, Filmmaking, E-commerce and Creative Production. These templates open with a ready-made workflow:
- Product shots for e-commerce
- Batch ad variation
- Cinematic brand trailer
- Social content studio
- Turn ideas into video
- Create dynamic product showcase videos
- Luxury jewelry campaign visuals
- Create brand packaging
Yes. Signed in, it saves to the cloud 5 seconds after your last change. Signed out, it stays on this device.
They use the same services but are priced from each model's base setting, so a node can cost more or less than the same model on a Create page. Read the price on the node's button.
Image and video generation nodes stop at four. A List node can hold up to 10 text items and 12 media items.