How to Edit Videos With ChatGPT (2026 Guide)

ChatGPT can now edit videos with Kapwing. We tested how to turn prompts into real edits, from subtitles to AI-generated footage.

How to Edit Videos With ChatGPT (2026 Guide)

In 2024, we tested whether ChatGPT could edit a video. Even a simple request, like trimming a clip or combining two files, usually ended with an error or instructions for doing the edit ourselves. ChatGPT could write scripts and suggest edits, but it couldn't reliably deliver a finished video.

That's changed with the introduction of the Kapwing MCP, which connects ChatGPT directly to a video editing workspace. You can now ask it to make the edits using Kapwing's tools. That includes uploading footage, adding subtitles, translating dialogue, generating B-roll, and exporting finished videos.

This is a game-changer, especially for new creators, who can now use natural language prompts to edit. Below, we'll show you how to connect ChatGPT to Kapwing, the exact prompts we used, and what you can realistically accomplish with AI video editing.

Table of Contents:

Can ChatGPT Edit Videos?

Yes, ChatGPT can edit videos when connected to a compatible video editor like Kapwing. ChatGPT doesn't have its own traditional video editing timeline, but it can use external editing tools to carry out instructions, make changes to projects, and return finished video files.

This is possible through the Model Context Protocol (MCP), an open standard that allows AI assistants to interact with external software. Think of it as giving ChatGPT access to the same editing tools you would normally use in a video editor, except you control them through conversation instead of clicking through menus.

With Kapwing's MCP integration, ChatGPT can help with several editing tasks:

  • Upload and organize footage: Import video files into your Kapwing workspace and create editable projects.
  • Add subtitles: Automatically transcribe speech and generate timed, word-by-word captions.
  • Translate videos: Convert subtitles into other languages while preserving their timing.
  • Dub audio: Translate spoken dialogue and generate audio using the original speaker's voice.
  • Generate video footage: Use Kai, to create new videos, images, and B-roll.

The distinction is that ChatGPT handles the instructions, while Kapwing processes the actual video. The resulting projects remain accessible in Kapwing's video editor, so you can review the output and adjust anything the AI gets wrong.

How to Edit Videos With ChatGPT Using Kapwing MCP

We'll walk through the individual tools first, then combine them into a single editing workflow.

The example prompts come from our tests in Claude Code, ChatGPT, and Codex, which can be adapted using the same integration.

Step 1: Connect Kapwing to ChatGPT

First, you need to give ChatGPT permission to use Kapwing's video editing tools. Open ChatGPT, navigate to Plugins and search for Kapwing. You can also add it directly here.

Next, connect your Kapwing account and authorize access to your workspace.

ChatGPT requesting permission to access Kapwing workspaces and use AI credits for video editing
Connect your Kapwing account and approve the permissions to get started.

I'll note that custom server setup and permissions depend on your ChatGPT plan and workspace settings. For example, some business workspaces require an administrator to approve external plugins before members can use them.

Once you're connected, start a new conversation and select the Kapwing plugin when you want ChatGPT to use its editing tools.

Kapwing plugin selected from the ChatGPT plugins menu for AI video editing
Once connected, select Kapwing from the plugins menu whenever you want ChatGPT to edit a video.

You can start with a basic request such as:

"Use Kapwing to find my most recent video project and give me a link to open it."

This is a useful way to verify the connection before attempting an edit.

ChatGPT finding a recent Kapwing video project and returning a link to open it in the editor
I started by asking ChatGPT to find my latest project. It returned the project details and a link to open it in Kapwing.

Step 2: Upload Your Video and Create a Project

Next, give the agent access to the video you want to edit. Depending on which AI agent you're using, you can provide a local file, a video URL, or a file already stored in your Kapwing workspace.

In our test, we placed a video in the working folder and asked the agent to upload it and create a project. Here's the exact prompt:

"Use the Kapwing MCP. Upload [CLIP NAME] from [FOLDER NAME] to my Kapwing Team Workspace and start a new project with it."
ChatGPT opening the Kapwing file uploader to import a MOV video into a team workspace
ChatGPT opened Kapwing's upload panel so I could send a video straight from my computer.

The agent uploaded the footage and created a new project inside Kapwing. We didn't need to open the editor and manually import the video, although the upload still needed time to process.

I'd recommend naming your project explicitly rather than saying something vague like "edit my last video." This is especially important when you have several similar exports or multiple projects in the same workspace.

Step 3: Add Automatic Subtitles With ChatGPT

Subtitles were one of the first editing features we tested. Instead of generating a transcript in ChatGPT and manually importing it into a caption editor, we asked the agent to create timed subtitles directly in Kapwing.

You can also specify how the captions should look, including their size, placement, and animation. This is the prompt that we used:

"Use the Kapwing MCP. In my Kapwing project Demo Clip, add word-by-word subtitles: big white text near the bottom of the frame, with the word being spoken highlighted in yellow."
0:00
/0:20

The agent added captions and highlighted each word as it was spoken, just as I'd requested.

Codex generated the subtitles and applied the requested word-by-word styling. The captions were synchronized to the original and placed directly in the project.

For precise visual styling, it's worth opening the project in Kapwing afterward. You can change fonts, adjust caption placement, or correct transcription mistakes without regenerating the entire video.

Step 4: Translate Video Subtitles Into Another Language

Once your video has subtitles, you can ask ChatGPT to translate the subtitles without having to export and re-upload a separate subtitle file.

Kapwing handles the translation and timing through its existing subtitle tools. That means the translated captions can stay aligned with the same sections of speech as the original. This is the prompt that we used:

"Use the Kapwing MCP. In my Kapwing project Demo Clip, translate the subtitles into Spanish on the same timings."
0:00
/0:20

Here's the same clip with Spanish subtitles, keeping the original timing and style.

This is particularly useful for tutorials, product demonstrations, and social videos where you want to reach audiences in languages without recording footage.

It's important to distinguish subtitle translation from dubbing. Translating subtitles changes the on-screen text but keeps the original audio. If you want viewers to hear your dialogue in another language, you'll need to use the dubbing tool instead.

Step 5: Dub Your Video Into Different Languages

Kapwing's AI dubbing tool goes a step further by translating the spoken audio itself. It can generate dialogue in another language using a recreation of the original speaker's voice.

Through MCP, you can ask an agent to create several language versions in one request.

For our test, we created five dubbed versions of the same clip: Spanish, French, German, Japanese, and Brazilian Portuguese. The instruction below comes directly from our combined editing test. This is the prompt that we used:

"Then dub that video into Spanish, French, German, Japanese and Portuguese (Brazil) in my own voice, one copy per language, each with captions in its own language."
0:00
/0:07

We also created five dubbed versions of the clip, each with translated speech and its own captions.

The advantage of requesting multiple languages together is that you don't need to repeat the workflow for each audience. The agent can create separate versions of the project, while Kapwing handles the translation, voice generation, and captions.

For the best results, start with footage that has clear speech and minimal overlapping dialogue. You should review each language version before publishing, particularly for names, technical terms, or expressions that don't translate literally.

Step 6: Generate New Video Footage With Kai

ChatGPT video editing isn't limited to modifying footage you've already recorded. Through Kapwing MCP, an AI agent can also send generation requests to Kai.

Kai can work with video and image generation, including Seedance, to create new footage based on a prompt or reference material. You can use this for B-roll, product shots, transitions, and scenes that would otherwise require filming.

We tested this by asking Kai to generate a five-second shot using Seedance 2.5, with a reference image of a laptop in a dark studio.

"Seedance 2.5, one single generation, 5 seconds, 16:9, with natural sound. A slow, smooth push-in towards the laptop on the desk in #image_1. Soft purple light, the timeline on the laptop screen glows gently, a faint room ambience and one soft keyboard tap. Keep the desk, the objects and the light exactly as they are in #image_1. Use #image_1 as a reference only, Seedance 2.5 reference-to-video: it is not the first frame. Make ONE single Seedance 2.5 generation of this whole prompt, no separate scenes and no assembly step."

For this prompt, #image_1 refers to an image attached as a reference in Kai. If you're generating from text alone, remove the instructions that refer to the image.

ChatGPT using the Kapwing plugin to request a five-second Seedance 2.5 AI video generation
I gave ChatGPT a video generation prompt, and it passed the request to Kapwing for processing.

Kai returned a generation plan with an estimated credit cost before producing the video. In this example, the five-second shot was quoted at 117 Kapwing credits and took approximately seven minutes to generate.

The result was saved in our Kapwing workspace, ready to review and use in a larger project.

0:00
/0:05

Here's the five-second shot generated from the prompt, complete with subtle camera movement.

We also generated a 20-second cinematic intro using a longer Seedance 2.5 prompt with seven distinct camera shots. That example demonstrates how MCP can handle more complex creative work, although longer generations require more careful prompting, reference images, and review.

For longer generations, I'd recommend creating a lower-resolution draft before committing to a full-quality render. During our test, the 480p draft cost 207 credits, followed by 185 credits for the full-res version and a 21-credit top-up.

0:00
/0:20

We also tested a more ambitious 20-second intro with multiple camera angles and cinematic sound.

Step 7: Combine the Editing Tasks Into One Prompt

The individual tools are useful, but the more interesting application is combining them. Instead of uploading a video, generating B-roll, adding captions, and exporting different language versions as separate manual tasks, you can describe the entire process in one request.

We tested this with Codex using a travel video. The goal was to upload the footage, generate two relevant B-roll clips, insert them into the video, dub the result into five languages, add subtitles, and export every version.

"Use the Kapwing MCP. Upload demo-raw-clip.mp4 from this folder to my Kapwing Team Workspace, using create_upload_url and a curl PUT. Then ask Kai to add 2 short B-roll shots generated with Seedance 2.5 over the lines where I describe a place, keeping my voice and my face on screen everywhere else. Then dub that video into Spanish, French, German, Japanese and Portuguese (Brazil) in my own voice, one copy per language, each with captions in its own language. After the dubs, add word-by-word English captions to the English version only. Then export all 6 versions and give me the editor link and the MP4 link for each one."

The create_upload_url and curl PUT instructions were specific to our Codex workflow. In ChatGPT, you can leave those implementation details out and tell the agent to use your existing Kapwing project or the video file available to it.

Kai planned the two B-roll shots at 199 credits in our run. We then received six exported versions: the English original and five dubbed videos, each with the appropriate captions.

0:00
/0:17

One combined editing request produced six video exports, each with its own language and captions.

You can also request more specific edits within the same workflow, such as changing the aspect ratio for TikTok, generating a particular style of B-roll, or keeping certain scenes untouched.

The more important your source footage is, the more specific your instructions should be. Name the project, identify which parts of the footage may be changed, describe the output you expect, and ask for the project links so you can review it.

The Results: Can ChatGPT Edit an Entire Video?

Our tests show that Kapwing MCP can complete a substantial video editing workflow using natural-language instructions. We went from source footage to a project containing generated B-roll, subtitles, translated audio, and six finished exports without manually completing each step on the timeline.

One of the biggest advantages was being able to continue working on the resulting videos in Kapwing. When the agent finishes an edit, it returns a normal project rather than a video that can no longer be modified. You can open the editor, correct captions, adjust the timing, replace a generated shot, or make other changes before exporting.

Kapwing video editor showing Spanish subtitles, an editable video timeline and AI-dubbed audio tracks
The finished videos are still editable. We opened the Spanish version in Kapwing to make a few final caption adjustments.

The process isn't completely hands-off. AI-generated footage still needs to be reviewed for visual consistency, subtitles may require corrections, and large requests can occasionally need a retry. Kai may also create a new project rather than overwrite an existing one, so always ask for the link to the latest result.

For the most reliable results, I recommend starting with one clearly defined task and confirming that it works before combining several edits into a longer prompt. Check your remaining Kapwing credits before generating expensive assets, and use the editor for the final visual adjustments.

Frequently Asked Questions

Can ChatGPT edit MP4 videos?

Yes. With a connected editor like Kapwing, ChatGPT can request changes to MP4 videos, including adding subtitles, translating dialogue, and creating new exports. Kapwing performs the video processing and provides an editable project or exported video file.

Can ChatGPT edit videos without a plugin?

ChatGPT can help plan edits, analyze supported media, and write code for video processing, but it doesn't provide the same dedicated editing timeline and export workflow as a connected editor. To carry out the Kapwing workflow described here, you need to give ChatGPT access to Kapwing's tools.

Is ChatGPT video editing free?

Kapwing MCP is available with free and paid Kapwing accounts, although AI features consume credits from your workspace. Free accounts have fewer credits and certain export restrictions. Your ability to use the Kapwing integration within ChatGPT also depends on your ChatGPT plan and workspace permissions.

There is no separate Kapwing MCP usage fee. You can review the current limits on Kapwing's pricing page.

Can ChatGPT add subtitles to a video automatically?

Yes. When ChatGPT is connected to Kapwing, you can ask it to transcribe your video and add timed subtitles. You can also describe caption styles in your prompt, although some formatting options may need to be adjusted manually in Kapwing's editor.

Can ChatGPT translate and dub videos?

Yes. The Kapwing integration supports subtitle translation and AI dubbing, including multilingual versions generated from a single request. In our test, we produced five dubbed language versions alongside the English original.

Use voice cloning only for your own voice or voices you have permission to reproduce, and review the translated dialogue for accuracy before publishing.

Can ChatGPT edit multiple clips together?

Kapwing MCP supports multi-asset project editing through Kai, including assembling footage, editing through a transcript, and positioning layers. These capabilities were added in the October 2026 MCP update.

For projects with several clips, give the agent clear instructions about their order and which footage to preserve. Open the resulting project to check transitions, audio continuity, and pacing before exporting.

Can I edit the video after ChatGPT finishes?

Yes. Videos created through Kapwing MCP are saved as editable Kapwing projects. You can open them in the browser to modify captions, replace footage, change timing, and make other adjustments without starting over.