How long can generated videos be?

It depends on the model: most produce clips between 4 and 10 seconds, and a few go up to 15. Each model's page lists its available clip lengths.

Can the AI generate sound with the video?

Yes, on models with native audio (like the Veo family) the clip includes generated dialogue, effects, or ambient sound. Models without audio produce silent clips. The model pages show which is which.

What happens if a generation fails?

You are not charged for failed generations. The reserved amount is returned to your balance automatically, and the chat shows what went wrong so you can adjust and retry.

Video generation

Describe a scene and the AI turns it into a short video clip, with motion, camera work, and on many models a generated soundtrack. Finished videos are saved to your workspace files automatically, so you can play, download, or share them later 🎬

Generate a video

Open a chat and describe the clip you want, for example "Make a 6-second video of waves crashing on a beach at sunset"
Mention the length, resolution, or model if you have a preference; otherwise the AI picks sensible defaults
Send the message. The video appears in the conversation with a player and a download button when it finishes

Video generation runs in the background and can take a few minutes depending on the model, length, and resolution. You can keep chatting or work elsewhere while it renders, and closing the tab does not lose it: the clip keeps rendering and appears in the conversation when it is ready. You can also cancel a generation while it is still running.

Start from an image

Many video models can animate a still image. Attach an image to your message and ask the AI to bring it to life. The image becomes the first frame of the clip.

This works great for turning generated images, product shots, or artwork into short animated scenes.

Generate an image first, then ask the AI to animate it, so you get precise control over the starting frame before spending on video.

Choose a video model

Different models trade off quality, speed, and price. Fast models are great for drafts and iterations; flagship models produce the most cinematic results with synchronized audio.

Browse the full catalog on the Video modelspage to compare per-second pricing, resolutions, clip lengths, and capabilities. Name the model in your message to use a specific one.

Where videos are saved

Every generated clip is automatically saved to a Generated Videos folder in your workspace files. You can find them in the file browser at any time.

You can ask for a different output folder or filename in your message to save the clip exactly where you want it.

Cost and plans

Video generation Usage Creditsneeds a subscription with a funded Credit balance: video bills against Usage Credits rather than your plan's included usage. There is no free video lane, because video is the most compute-intensive thing a model can make.

Pricing is per second of output video and varies by model, resolution, and audio. The cheapest rate is around $0.05 per second at 720p, and the shortest clip is 4 seconds, so even a short, low-cost clip starts around $0.20 to $0.25 rather than a few cents. Flagship models at high resolution cost a few dollars. The model pages show exact per-second rates.

Keep drafts cheap: iterate with a fast, low-cost model at 720p, then re-run your best prompt on a flagship model for the final cut.

FAQs

Image generation

Ask AI to create images from text descriptions, saved to your Drive automatically.

Audio generation

Turn text into natural-sounding speech, with a voice and model you choose.

Chat settings and toggles

Customize each conversation from the composer: tool toggles, autonomy, compaction, and a spend limit.

Up next

Talking to idapt with your voice

Dictate a message, hold a full spoken conversation, and hear any reply read back to you.

Was this helpful?