ChatGPT has no video model of its own, but it can make AI videos once you add the Animx plugin. Type @Animx in a chat, describe the shot, and the finished clip comes back in the same conversation.
This guide covers setup, what to put in a request, which model to choose and what each render costs.
Why ChatGPT needs a plugin for video
OpenAI closed the Sora app and website on April 26, 2026, and the Sora API on September 24, 2026. Video in ChatGPT now comes from plugins: ChatGPT handles the conversation, and the plugin’s service renders the clip.
That split suits video work. You can shape an idea with ChatGPT, then send the shot you like to Animx without retyping the brief in another tool. Animx puts 40+ video and image models on one subscription, so you can switch models without switching accounts.
How to turn ChatGPT into a video generator
You need a ChatGPT account and an Animx account. New to Animx? Sign up free for 200 credits, no card needed. There is no API key or developer setup.
- Open the Animx plugin page in ChatGPT, or go to Plugins in ChatGPT and search for Animx.
- Select Install plugin. When asked, select Connect and sign in with your Animx account.
- Start a chat, type @Animx and describe what you want.
A first request can be this simple:
@Animx 5-second product video: perfume bottle on dark rocks at sunset, slow orbit, warm rim light
Every render spends Animx credits, and the 200 free credits won’t cover a Seedance 2.5 clip. Check the costs below before your first video.
Can’t find Animx in the directory? OpenAI says access depends on your plan, workspace, role and region. On a company workspace, ask your admin to allow the plugin.
What to put in your request
Only the prompt is required. Name any other setting when it matters; otherwise the plugin uses its default.
| Setting | What to write | If you leave it out |
|---|---|---|
| Prompt | Subject, action, camera move, light and mood | Required |
| Model | “Use Seedance 2.5” | Animx’s default model |
| Format | 16:9, 9:16 or 1:1, if the model supports it | 16:9 |
| Length | 2 to 30 seconds, depending on the model | 8 seconds |
| Resolution | 480p up to 4K, depending on the model | Varies by model |
| Image | A public HTTPS link, used as the first frame or as a reference (up to 4, on models that support them) | Text only |
Write the prompt like a shot list. Lead with the subject and action, then add the camera move, light and mood. Aim for 20 to 60 words; the examples below are shorter so they’re easy to adapt.
ChatGPT can draft this part too. Ask for a few ideas, pick one, then send it to Animx:
Give me three 30-word shot ideas for a coffee brand ad.
@Animx make the second one as a 10-second 16:9 video.
If your exact wording matters, tell ChatGPT to pass your prompt to Animx word for word.
Which model to ask for
Choose by the shot. Credits below are for a 5-second clip at 720p; longer or higher-resolution clips cost more.
| Model | Pick it for | Sound | Credits |
|---|---|---|---|
| Wan 2.7 | Directed camera moves, up to 15 s | Yes | 300 |
| Happy Horse | Lip-synced speech in 7 languages | Yes | 420 |
| Kling 3.0 | Natural real-world motion, up to 15 s | No | 505 |
| Seedance 2.5 | References, start and end frames, takes up to 30 s | Yes | About 554 |
| Veo 3.1 | Photoreal scenes with dialogue, up to 8 s | Yes | 1,200 |
| Gemini Omni Flash | Cinematic realism with native audio, 4 to 8 s | Yes | Shown in the generator |
For cheap drafts, Veo 3.1 Lite costs 150 credits and Seedance 2.0 Mini costs 228 for the same clip. The lineup changes, so ask for the current list:
@Animx which video models can I use? Two requests to try
A vertical clip from one line
@Animx use Seedance 2.5, 9:16, 15 seconds: woman in a red dress turns to the camera between marble columns, golden hour, soft wind
This prompt comes from our Seedance 2.5 page, where you can watch the clip it produced. Naming the model, format and length leaves nothing to a default, and Seedance 2.5 generates sound with the picture.
Cost: a 5-second clip at 720p is about 554 credits, and a 15-second take costs more.
A product video that starts from a still
The plugin makes images too, and an image costs far less than a video, so test the look first. GPT Image 2 costs 30 credits at 1K, so you can try 18 stills for the price of one 5-second Seedance 2.5 clip.
@Animx make a 16:9 image with GPT Image 2: perfume bottle on dark rocks at sunset, warm rim light
When you like the still, animate it:
@Animx use that image as the first frame. Seedance 2.5, 5 seconds, 720p: slow orbit around the bottle, warm rim light
Cost: about 584 credits for one still and one clip.
When the video is ready
Renders usually take 30 to 90 seconds, or longer when the queue is busy. The finished video appears on a card in the chat, with its download link.
To check on a render, ask “check my last generation.” To change something, say so in the same thread, such as “same shot, slower orbit.” Each new version is a new generation and spends credits again.
What it costs
Every generation uses the credits of your signed-in Animx account, the same as in the Animx app. A ChatGPT subscription doesn’t pay for rendering.
| Plan | Price per month | Credits | 5-second Seedance 2.5 clips at 720p |
|---|---|---|---|
| Free | $0 | 200, one time | Not enough for one |
| Basic | $5 | 700 a month | About 1 |
| Plus | $39 billed annually, $49 monthly | 10,000 a month | About 18 |
| Ultra | $99 billed annually, $129 monthly | 30,000 a month | About 54 |
Basic includes selected entry models plus Seedance 2.5. Plus and Ultra include all models. A top-up of 5,000 credits costs $30, and those credits never expire.
To avoid surprises, start the thread with this:
Before you generate anything, tell me the model, length and resolution, then wait for my OK.
To see the exact credit cost of a longer clip, set it up in the Animx generator, which shows the price before you click Generate.
Questions about Animx in ChatGPT
Is the Animx plugin free?
The plugin works with free and paid accounts, but every render spends credits. New accounts get 200 free credits, which is less than one Seedance 2.5 clip.
How long can a video be?
From 2 to 30 seconds, depending on the model. Seedance 2.5 goes up to 30 seconds in one take, Kling 3.0 and Wan 2.7 up to 15, and Veo 3.1 up to 8.
Can I use my own photos?
Yes, if the image has a public HTTPS link. It can be the first frame, and models that support references accept up to 4 through the plugin. For more, the Animx site takes up to 9 on Seedance 2.5.
What if a render gets stuck?
Ask ChatGPT to check your last generation, which reports its queue position and progress. If a render is never picked up, it’s marked as failed after 25 minutes and your credits return automatically.
Can I use the videos commercially?
Yes. You own what you make, and every paid plan includes commercial rights.
Does Animx work in Claude too?
Yes. In Claude, Animx connects as a custom connector rather than a plugin. Setup steps are in the Animx docs.
Ready to try it? Add Animx to ChatGPT. New accounts get 200 free credits, no card needed.
