Generate images, videos and voice with AI

In short: open Generate with AI, pick what to create, describe it and check the price. You see the total before you generate. If the generation fails or the model refuses the request, your credits come back on their own.

How to generate

1. Open Generate with AI.
2. In What to create, pick Image, Video or Voice.
3. Describe what you want to see. For voice, write the text the voice will read.
4. Pick the quality and the settings.
5. Check the price and generate.

The result shows up in Your creations as soon as it's ready. You don't need to keep the screen open.

The request text takes up to 2,000 characters. Stuck? Use Improve: the AI rewrites your request for 1 credit. If it doesn't work, the credit comes back.

Starting from images

You can send images along with the request:

• Image: to edit a photo or use it as a reference.
• Video: in Use the images as, pick First scene (the video starts on that image) or Reference (the model uses the person, product or style from the image in a new scene).

Each model takes a certain number of images. The screen tells you when a model doesn't fit your request.

Quality and model

The quickest way is to pick a preset: Fast, Balanced or Best quality.

To choose yourself, tap Choose a model. The list goes from cheapest to most expensive, already at the resolution and sound you picked. Each model shows a quality score from 0 to 100.

The settings change the price:

• Length (video): every second is charged. Start short to test the idea.
• Resolution (video): sharper costs more.
• Generate with sound (video): adds voice, music and background sound to the video.
• Quantity (image): each image comes out different and is charged separately.

What it costs

The price depends on the model:

• Image: per image.
• Video: per second.
• Voice: per 1,000 characters, charged on the exact number of characters in your text.

The screen shows the total in credits, and the approximate amount in money, before you generate. If the price changes in the meantime, the screen shows the new amount and nothing is charged until you generate again.

If you're short on credits, you can buy them right there. After the purchase, the generation starts on its own. In a team workspace, ask the workspace owner to add credits: your choices stay on the screen.

Up to 4 at once

Each workspace runs up to 4 generations at the same time. With 4 running, wait for one to finish before starting another.

When credits come back

They come back on their own when the generation fails or the model refuses the request. For example: a protected work or brand, a real face shown too prominently, or a request blocked by the model's content rules.

If the result came out with a flaw, tap Request a refund, pick the reason and tell us what happened. You get an answer within 24 hours.

You can't ask for a refund after you Download the file or after you use the video in a scheduled or published post. If the post hasn't gone out yet, cancel the schedule and ask again.

After generating

In Your creations, you can download, use Edit and caption to open the video in the editor, or tap Publish.

What about music, sound effects and 3D?

They're not on this screen. Today they come through the API and MCP. See API and MCP for developers.

For a full narrated video from a topic, see AI narrated stories.