Can ChatGPT make product videos?
Yes. Install the FLORA app in ChatGPT and it can turn a product image into a short video with a model such as Seedance 2.5. Give it a product shot, describe the camera move, and it returns an MP4. In our run, a 5-second vertical 720p product clip cost $2.457 and took 3 minutes 40 seconds through FLORA.
Key takeaways
- ChatGPT makes product videos once the FLORA app is installed from flora.ai/mcp.
- Our 5-second 9:16 product clip with Seedance 2.5 cost $2.457 and took 3 minutes 40 seconds.
- The product shot it started from cost $0.055 and took 30 seconds with GPT Image 2.5 Flare.
- A slow push in kept the bottle's shape and color from the first frame to the last.
- Aspect ratio auto kept the vertical 720 x 1280 frame of the product shot.
At a glance
| What it does | Animates a product image into a short video clip |
|---|---|
| Works in | ChatGPT with the FLORA app, and other agents with the FLORA MCP |
| Model | Seedance 2.5 image-to-video, id i2v-gengateway-seedance-2-5-i2v |
| Inputs | One product image and a motion prompt |
| Outputs | MP4, 4 to 30 seconds, vertical, square or wide, with audio |
| Typical cost | $2.457 for 5 seconds at 720p |
| Typical time | 3 to 4 minutes per 5-second clip |
What you need
- ChatGPT with the FLORA app installed. Install steps are on the FLORA MCP page.
- A FLORA account on a workspace with video generation.
- A product image with a public link, or one made in the same chat.
Key features
- Product stays put: Image-to-video starts from your product shot, so the product is fixed before motion.
- Vertical by default: Aspect ratio auto keeps the ratio of the shot, so a 9:16 still gives a 9:16 clip.
- Quote first: FLORA returns the price in dollars before a clip runs.
- Canvas history: Each clip is saved on a FLORA canvas beside the product shot it came from.
Best for
- Short product loops for Reels, TikTok and Stories
- Product page videos from an existing packshot
- Shops testing motion before booking a video shoot
- Launch teams who already draft copy and briefs in ChatGPT
Step by step
Step 1: Install FLORA in ChatGPT
Open flora.ai/mcp, install the FLORA app in ChatGPT and sign in to FLORA.
Step 2: Give it the product shot
Paste a public link to the product image, or ask ChatGPT to make one first. Use the ratio you want the video in.
Step 3: Ask for one camera move
Product clips work best with one slow move and a line that says the product must not change.
Animate this product shot with Seedance 2.5 through FLORA, 5 seconds, 720p, aspect ratio auto. Slow push in toward the bottle, steady and smooth. A light breeze moves the pine needles, thin mist drifts over the lake. The bottle stays exactly the same shape and color, no text appears. Keep the light soft and overcast, no lens flare. Quote it first.Step 4: Check the product, then download
Watch the first and last frames for changes to the product: shape, color, cap, any print. Rerun with a sharper note if anything moved.
What happened when ChatGPT made a product video with FLORA?
On October 16, 2026, we ran this from an agent session through the FLORA MCP, the same FLORA tools the ChatGPT app calls. We had no product of our own, so the agent first made the product shot with t2i-gpt-image-2-5-flare at 9:16: a matte olive-green insulated bottle with a brushed steel cap on a granite ledge, pine needles beside it, a lake and mountains behind. It took 30 seconds and cost $0.055.
A quote call returned $2.483. The agent then called flora_create_generations with model i2v-gengateway-seedance-2-5-i2v, the shot in params.image_url, duration 5, resolution 720p and aspect_ratio auto, plus the push-in prompt. It finished in 3 minutes 40 seconds and was charged $2.457.
The clip is 5.06 seconds, 720 x 1280 at 24 fps, with a quiet ambient audio track. The camera pushes in slowly and steadily. The bottle's shape, olive color and steel cap match the first frame in every second we checked.
What went wrong: our prompt said the morning light slowly brightens, and the model read that as a sunrise. By the last second the background turns warm gold with a lens flare on the bottle, which is a different mood from the soft shot we started with. The fix is to leave light changes out of a product prompt, or say "light stays soft and overcast, no lens flare", as the prompt in step 3 does.
- Models
- t2i-gpt-image-2-5-flare, i2v-gengateway-seedance-2-5-i2v
- Cost
- $0.055 shot, $2.457 clip ($2.483 quoted)
- Time
- 30 s shot, 3 min 40 s clip
- Output
- MP4, 5.06 s, 720 x 1280, 24 fps, with audio
Push in or orbit: pros and cons
Pros
- A slow push in kept our bottle identical across the whole clip.
- A locked-off shot is the safest choice for labels and fine print.
- An orbit shows the back of the product, which a still cannot.
Cons
- A push in shows nothing new about the product.
- An orbit asks the model to invent the sides it has never seen, so check them closely.
- Any video run costs dollars, so settle the product shot first.
Does ChatGPT render the video itself?
ChatGPT plans the request and calls FLORA through the app. The video model in FLORA renders the clip, and ChatGPT returns the link. The clip also lands on your FLORA canvas, where you can try another take from the same product shot.
Can I start from a real product photo?
Yes, and it is the better start for a real product. Paste a public link to the packshot. If the photo is cluttered, ask for a clean product shot made from it with an image-to-image model first, then animate that.
Which camera moves suit products?
Pick one move per clip and keep it slow.
- Locked off, with something small moving around the product.
- Slow push in toward the product.
- Slow pan across a lineup of products.
- A turntable spin, only when the product's other sides are simple.
Frequently asked questions
How much does a product video cost?
Our 5-second 720p Seedance 2.5 clip cost $2.457. The product shot we started from cost $0.055. FLORA quotes each run in dollars before it starts.
Can I get vertical video for Reels and TikTok?
Yes. Start from a 9:16 shot and set aspect ratio to auto, or ask for 9:16 directly. Our clip came back at 720 x 1280.
Will the product change shape?
It did not in our clip. A slow, simple camera move and a line saying the product must not change help keep it that way. Always check the last frame.
Where do I install FLORA in ChatGPT?
From flora.ai/mcp, which links to the FLORA app for ChatGPT and walks through sign-in.