Video editing guide
ChatGPT video editor: edit videos with AI prompts
Upload a video, describe the change you want and review the edit on a timeline. GPTVideoEditor lets you request cuts, timed text, audio changes and a different frame shape, then export the result in your browser.
How a ChatGPT video editor turns a prompt into an edit
GPTVideoEditor combines a browser video editor with an AI agent. Upload an existing video, describe the change you want and review the result on an editable timeline. The agent can turn an instruction such as “Add Welcome at the bottom for the first three seconds” into a timed text layer. You can then change that text, adjust its timing or undo the edit before exporting.
The model plans editing operations; your browser applies them to your project and renders the finished file. A prompt does not replace your video with newly generated footage. Ordinary cuts, text, volume changes and framing adjustments work with the source you uploaded. This makes the difference between requesting an edit and generating a new video important: this editor works on existing footage.
GPTVideoEditor is an independent service, not an official ChatGPT product. Quick and Standard use OpenAI language models, while Advanced uses an Anthropic model. The conversational interface is a way to control the editor. Your ChatGPT subscription does not supply credits here, and using this site does not require a ChatGPT account.
How to edit a video with AI prompts
Start with a copy of the original and a clear goal. A short clip with one requested change is easier to check than a long recording with several vague instructions. Manual timeline editing and browser exports do not spend AI credits; sign in when you want to use the agent.
Upload your source video
Drop a video into the upload box or browse for a file on your device. The beta accepts videos up to 1 GB and projects up to 15 minutes. Import depends on whether your browser can decode the video and audio codecs. H.264 video with AAC audio in an MP4 is a useful starting format when another file fails to open.
Describe a specific change and its timing
Name the action, the target and the time range. “Mute the sound from 00:08 to 00:12” is clearer than “fix the audio.” For text, include the exact words, position and duration. Select a clip when the instruction concerns its volume, speed or framing.
Choose the model and context
Select Quick, Standard or Advanced. The request displays a maximum credit reservation, with unused reserved credits returned after completion. Include frames only when visual context is useful: that option shares sparse image samples, not the entire video or a complete record of every scene.
Review the timeline and preview
Play the changed section, inspect the text and listen to the sound. Confirm that the agent edited the intended clip and time range. Use the timeline controls for a precise correction or undo a change before trying a narrower prompt.
Export the finished video
Choose MP4 or WebM and the available maximum output height. The editor does not upscale your source just because you choose 1080p. Keep the tab open while the browser encodes the video, then open the downloaded file to check both the picture and the sound.
How to write a useful ChatGPT video editing prompt
A useful video editing prompt has four parts: the action, the target, the timing and what should stay unchanged. The target can be the selected clip, a named text layer or a range in the current project. Use the timeline you are looking at now. After removing footage, later timestamps move earlier, so an old timestamp may no longer identify the same moment.
Separate changes that need different kinds of judgment. First remove a clearly defined pause, then decide whether the next cut needs a fade. First add the exact title, then adjust its size after watching the preview. This keeps the reason for each change visible and gives you a useful point to return to if the next request goes wrong.
For example: “Add the text Product demo at the bottom from 00:00 to 00:03. Use white text with a dark background. Keep the clips and audio unchanged.” That request supplies wording, placement, timing and boundaries. “Make the video professional” leaves the agent to guess which of those choices matters.
Video editing prompts for cuts, text, audio and framing
Use these examples as starting points and replace the times with positions from your current timeline. Watch the full changed section afterward, including a little footage before and after it. A correct operation can still produce an awkward cut when it lands in the middle of a gesture or sentence.
Add a title to the intro
“Add the exact text Welcome to the tutorial at the bottom from 00:00 to 00:03, in white with a dark background.” Check spelling, readability and whether the title covers a face, button or other important detail.
Remove a known unwanted section
“Remove the section from 00:12 to 00:15 and join the remaining footage.” Check the join for missing words or a sudden change in movement. The project becomes shorter after that range is removed.
Make jump cuts around long pauses
“Remove the long pauses from this video. Keep the spoken content.” Pause requests can use local audio silence analysis. A quiet section is not always an unwanted pause, so listen to each join and restore deliberate pauses when needed.
Increase a selected clip’s volume
“Set the selected clip’s volume to 150%. Leave its speed and framing unchanged.” Listen for distortion: increasing gain raises background noise too and cannot reliably repair an already clipped recording.
Mute a specific time range
“Mute the audio from 00:08 to 00:12 without removing the footage.” Check that audio returns at the right moment. Muting a range affects the sound in that range; it does not selectively remove just one voice or sound source.
Make a vertical version
“Make this video 9:16 with a centered crop.” Check the subject throughout the clip. A centered crop can cut off someone who moves near an edge; it is not automatic subject tracking.
Change playback speed
“Play the selected clip at 1.5 times its normal speed.” Check both the shorter duration and the sound. This beta changes audio pitch with playback speed, so it is not a pitch-preserving voice adjustment.
Add a fade
“Add a 0.6-second fade to the selected clip.” The available fade passes through black and changes the sound at the clip edges. It is not a cross-dissolve that overlaps two videos.
Review AI video edits with the browser timeline
The timeline remains the source of truth after an AI request. A cut changes clip boundaries or removes a range; a text request creates a layer with a start and end; a sound request changes gain or adds a mute interval. Review those changes rather than relying only on the agent’s written answer. Undo gives you a way to compare with the previous project state.
Manual controls are useful when you already know the exact edit. Split a clip at a chosen point, shorten an unwanted opening or adjust a text layer without spending another AI credit. The agent is useful for translating a written instruction into those same editing operations. Both methods work on the same project, so you can use a prompt for the first pass and make a precise correction yourself. With nothing selected, the right panel shows your media library. Import video or audio, then drag it onto a compatible track or use its + button. Add another video track for an overlay or picture-in-picture; separate audio tracks let you arrange music and voice recordings independently. Deleting a clip leaves a gap; removing a time range joins the remaining project across all tracks.
What the AI agent can understand about your video
The agent receives project information such as clip durations, ordering, text layers and the selected clip. When you enable Include frames, it also receives a small set of sampled images. Those samples may help with a visible title or a broad description of a scene, but they can miss a short event between samples. A filename or one frame is not evidence of everything that happens in a recording.
For a reliable edit, provide the location yourself when possible: “Remove 00:17 to 00:20” is more specific than “remove the part where the person waves.” Silence analysis can identify low-audio ranges for pause requests, but it does not produce a full speech transcript. Automatic transcription and captions, object removal, generated footage and advanced video effects are not included in this beta.
What to check when an AI video edit goes wrong
If the wrong clip changed, undo and select the intended clip before repeating the request. If a time range changed unexpectedly, check whether an earlier cut shortened the project and moved later footage. Read the time on the current timeline instead of reusing a timestamp from the original recording.
If text is missing, move the playhead inside the layer’s start and end times. Check that its color contrasts with the background and that the words fit the frame. If sound disappears, inspect the clip volume and mute ranges, then play a section outside the affected interval to find where it returns.
If import fails, the file extension alone does not establish codec support. An MP4 can contain codecs your browser cannot decode. Try a compatible H.264/AAC MP4 and confirm that the source plays correctly before importing. If export fails, keep the tab active, reduce the output height or shorten the project. Device memory and encoding support affect what your browser can finish.
Export MP4 or WebM and keep your original video
MP4 export uses H.264 video and AAC audio when your browser can encode them. WebM uses VP9 and Opus when supported. The selected maximum output height is a ceiling, not a promise to create new detail. Check the final video in the player or publishing service where it will be used, particularly if you changed its aspect ratio.
Encoding happens in your browser. Supported browsers can write a longer export directly to a chosen file. The fallback that prepares a download in memory is limited to five-minute projects. A browser tab that can preview a video is not automatically able to encode every output format, so available codecs and device memory matter.
Projects and originals can be kept in browser storage on the current device. A project backup contains editing instructions and file identities, not the original footage. Saving project edits to your account also saves metadata rather than uploading the full original video. Keep the source files separately; clearing browser data can remove locally stored originals.
AI video editing models, credits and pricing
Quick, Standard and Advanced control which model plans the edit. They use the same timeline tools and browser export engine. Choosing a more expensive model does not change the resolution limit or unlock effects the editor cannot perform. Use the request estimate to judge the cost of the specific instruction and context you are sending.
Manual edits and browser exports do not spend AI credits. Paid plans provide a monthly AI allowance; annual plans are paid up front but still refresh their credits monthly. Included credits do not roll over. Purchased extra credits carry over, and an active video subscription is required to buy them. Video plans and balances are separate from GPTPhotoEditor’s image plans, even when you use the same sign-in credentials.
Frequently asked questions
What is a ChatGPT video editor?
In this service, it means a video editor controlled with ChatGPT-style written instructions. GPTVideoEditor uses AI language models to plan supported changes to an uploaded video, then applies those changes to an editable browser timeline. It is independent of ChatGPT and OpenAI.
Can I edit a video I upload?
Yes. Import an existing video, use manual controls or ask the AI agent for an edit, review the project and export a new file. This beta supports projects up to 15 minutes and source files up to 1 GB, subject to browser codec support.
Can I add text to the beginning of a video?
Yes. Specify the exact wording, position and time range, such as white text at the bottom from 00:00 to 00:03. Timed text is rendered into the exported video. Review the preview to confirm readability and spelling.
Can AI make jump cuts or remove pauses?
The agent can remove ranges and use locally detected silence ranges for pause requests. Silence does not prove a section should be removed. Listen to the cut points and undo changes that remove deliberate pauses or affect speech.
Can I mute part of a video without cutting it?
Yes. Ask to mute a time range while keeping its footage. Muting the selected clip’s volume affects that clip’s sound; a timed mute affects the specified project interval.
Does it generate new videos or remove objects?
No. The current editor works with existing footage and supported timeline operations. Generated footage, object removal, automatic transcription and captions, and advanced generative video effects are not included.
Is online video editing free?
Manual editing and browser export do not spend AI credits. AI requests use this service’s credit balance and paid plans. A ChatGPT subscription does not provide GPTVideoEditor credits. The pricing page lists the current video plans.
Does my whole video get uploaded for an AI request?
Ordinary editing and export happen locally. AI requests send the prompt and project context. Enabling Include frames additionally shares sparse sampled images. Saving project edits to an account stores metadata, not the full original video.
Can I export a vertical video for social media?
Yes. Choose a 9:16 project shape and check the framing. Fit keeps the full source frame with space around it; a centered fill crops the source to cover the frame. The editor does not automatically track a moving subject.
Which files can I export?
Supported browsers can export MP4 with H.264/AAC or WebM with VP9/Opus. Available encoders and device memory determine whether a particular export can complete. Output height is limited by the source and selected setting.
Video editing guides by task
Follow a complete workflow for the change you need, with example prompts, review steps and advice for the exported file.
Write ChatGPT-style video editing prompts with exact timestamps, clip targets and clear instructions. Examples for titles, jump cuts, muting and resizing.
Trim a video in your browser, cut unwanted sections and review the join before exporting. Follow precise manual and AI prompt workflows.
Add titles and text overlays to an uploaded video. Set the words, position, size and time range, then render them into an MP4 or WebM export.
Use local silence analysis and AI prompts to cut long pauses from a video. Review speech boundaries, deliberate pauses and each join before exporting.
Mute an entire clip or a specific video time range without removing footage. Adjust volume, review audio boundaries and export with the changes applied.
Change a video to 9:16, 1:1 or 16:9 in your browser. Choose fit or centered crop, check moving subjects and export without unnecessary upscaling.