
How to Use ChatGPT Voice Mode: A Simple Guide (2026)
Learn how to use ChatGPT voice mode on your phone and PC. Free limits, Live vs Advanced, voice choices, and fixes when it stops working.

Can ChatGPT make videos? The short answer is no โ not on its own. ChatGPT cannot render a video file the way it writes text or makes a still image. But that is not the whole story. With the right add-on, or with ChatGPT doing the planning and another tool doing the rendering, you can go from an idea to a finished clip without ever leaving the chat window.
This guide gives you the honest answer as of November 2026: what ChatGPT can and cannot do with video, the two methods that actually work, what each one costs, and the limits you will run into โ so you do not waste time on a method that was shut down months ago.
No. As of November 2026, ChatGPT has no built-in video generator. If you type “make me a video of a dog running on a beach” into ChatGPT, you will not get an MP4 file back. What you can get is a script, a shot list, a storyboard, captions, or a still image โ all useful, but none of them is a video.
There are two ways to get a real video out of the process:
Both methods work. Which one fits you depends on your budget, how long the video needs to be, and whether you want to stay inside ChatGPT or not.
If you remember Sora, you are not imagining things. Sora was OpenAI’s own text-to-video model, and for a while it lived inside ChatGPT. OpenAI closed the Sora app in April 2026, and the Sora API followed in September 2026. That is why so many older articles and videos say “just use Sora in ChatGPT” โ that advice is out of date.
For context, when Sora was still available, Plus users could make about 50 videos a month at 720p and up to 10 seconds each, while Pro users got around 500 videos a month at 1080p and up to 20 seconds. None of that applies anymore. If a tutorial tells you to open Sora inside ChatGPT, close it and find a newer one.
ChatGPT cannot render video, but it is genuinely useful at every step around it:
Think of ChatGPT as the writer and director, not the camera. It plans the video; something else shoots it.

This is the closest thing to “ChatGPT makes a video.” You install a video app from the ChatGPT marketplace, and it connects ChatGPT to a video model. The video is rendered by that model, not by ChatGPT itself, but the whole thing happens in the chat.
Popular options include Atlas Cloud, getimg.ai, Pictory, Synthesia, and Toybox. Each one connects to different video models โ Seedance, Veo, Kling, and others โ and each has its own prices and limits.
Here is an example of a prompt that works well:
“Make a 6-second video, 16:9, of a golden retriever puppy running along a beach at sunset. Slow camera follow from the side, warm light, realistic style, with the sound of waves.”
The details matter. Subject, action, setting, camera, light, style, length, shape, and sound โ the more of these you include, the closer the result will be to what you pictured.
Most clips from these plugins are short: about 4 to 30 seconds each, depending on the model. That is a limit of the video models, not of ChatGPT. If you need a longer video, you make several short clips and join them in a video editor, or you use one clip as the starting frame for the next one.
Avatar and presentation tools are different. Synthesia, for example, can make talking-head videos up to 60 minutes on paid plans (about 3 minutes free) at 1080p, because it animates a presenter reading a script instead of generating new footage frame by frame.
It is not free, and it is not part of your ChatGPT subscription. Each clip costs credits from the video tool. As an example, one tested plugin charged about $1.82 for a 6-second clip and about $2.42 for an 8-second clip at 720p. Prices change often, so check the tool’s own pricing page before you start.
Some tools offer a small free tier. Pictory’s free plugin, for example, allows a few short videos a month at 720p, and Toybox offers free 10-second clips at 480p with a watermark. These are fine for testing, but not for regular use.
A quick bit of math helps: at roughly $2 per 6-second clip, a 1-minute video made of ten clips costs around $20 in credits, plus your time joining them together. That is still far cheaper than hiring a video crew, but it is not “free.”
No ChatGPT plan includes video generation. The plugins are separate services with their own accounts and prices. Whether a plugin works on a free ChatGPT account depends on the plugin โ some have been tested on free accounts, while others say they need Plus. If you are on the free plan, check the plugin’s own page before you pay for anything.
This is the method most people use, and it costs little or nothing. ChatGPT does the writing, and a dedicated video tool does the rendering.
This method takes more steps, but it has real advantages: you can use the free tier of the video tool, you can pick the best model for each scene, and you are not locked into one plugin’s prices.

Yes โ with a plugin, or with a separate video tool. This is called image-to-video, and it is one of the most reliable ways to get a good result.
The idea is simple: you give the tool a photo, and it animates it. A product photo can slowly rotate. A portrait can turn its head. A landscape photo can get moving clouds and water. Because the tool starts from a real image, the result usually looks more accurate than a video made from text alone.
To do it inside ChatGPT, attach the photo to your message when you prompt the video plugin, and say what should move. For example: “Animate this photo so the camera slowly zooms in and the trees sway gently, 6 seconds, 16:9.”
Two things to know. First, faces and hands are still the hardest part โ a face can shift or change slightly during the clip, so keep movements small. Second, do not animate photos of real people without their permission, and never use someone else’s photo as if it were yours.
Yes, in the same way: ChatGPT turns your text into a detailed prompt, and a video model renders it. This is called text-to-video. You describe the scene in words, and the model creates the footage from scratch.
Text-to-video works best for simple scenes with one clear subject and one clear action: “a red bicycle leaning against a brick wall in the rain” will do better than “a busy market street with ten people talking.” Keep the scene simple, and the model has less to get wrong.
For the best results, give the model reference images when you can. If your video has a character or a product that must look the same in every scene, attach a photo of it and tell the tool to keep it consistent. Without a reference, the same character can look different from one clip to the next.
It helps to know the limits before you start, so you do not waste credits on something that was never going to work:
In most cases, yes โ with two conditions. First, check the rights on the tool you used. Most video tools give you commercial rights to videos you generate on a paid plan, but free tiers sometimes limit you to personal use. Read the tool’s terms before you use a clip in an ad or a product video.
Second, do not use AI video to mislead people. A video that shows a real person saying or doing something they never did can break the law in many US states, and platforms like YouTube and TikTok require you to label AI-generated content. If your video shows a real person, a real product, or a real event, say so clearly, and get permission when a real person’s face or voice is involved.
As a rule: if you would not do it with a photo, do not do it with a video either.
Partly. Writing scripts, prompts, and shot lists in ChatGPT is free, and most video tools offer a small free tier โ a few short, watermarked clips a month. But a steady supply of good clips costs money, either in credits (roughly $2 per short clip) or a subscription to the video tool.
ChatGPT itself makes no video, so there is no ChatGPT limit. The video models behind the plugins usually make clips of 4 to 30 seconds each. For anything longer, you join several clips in an editor. Avatar tools like Synthesia are the exception โ they can make talking-head videos up to an hour long.
Yes. Attach a photo to your message and ask a video plugin to animate it, or upload the same photo to a video tool like Veo or Kling and choose image-to-video. The tool animates the photo โ a slow zoom, moving water, a turning head โ instead of creating the scene from scratch.
Yes, through a plugin or a separate tool. You describe the scene in words, ChatGPT turns it into a detailed prompt, and a video model renders the clip. Keep scenes simple โ one subject, one action โ for the best results.
There is no single best tool. Seedance is often rated highest for realistic, cinematic clips, and Kling is often picked as the best value. Veo is strong for realistic motion, Runway is popular with creators, and Synthesia is best for talking-head and training videos. Try the free tier of two or three and compare them with the same prompt.
No. OpenAI closed the Sora app in April 2026 and the Sora API in September 2026. Old guides that tell you to use Sora inside ChatGPT are out of date. Video generation in ChatGPT now works through third-party plugins instead.
Last updated: November 2026. Video tools, prices, and ChatGPT features change often โ check the tool’s own site for the latest details before you pay for anything.

Learn how to use ChatGPT voice mode on your phone and PC. Free limits, Live vs Advanced, voice choices, and fixes when it stops working.

Learn how to use ChatGPT for resume writing with copy-paste prompts, an ATS self-check, and before-and-after bullet examples.

A side-by-side look at ChatGPT Free and Plus โ price, models, and every limit that matters.