RendGet started

Blog · November 17, 2026

ChatGPT Video Editing: Connect an Editor and Make Real Videos

ChatGPT can edit videos once you connect it to a video editing MCP server. By itself it can describe an edit, write a script or suggest an FFmpeg command, but it doesn't own a timeline or a renderer. Add a server like Rend (https://rend.ai/api/mcp) as a custom MCP server in ChatGPT, sign in, and ChatGPT can cut your footage, add captions, voice and music, and hand back an MP4 plus a timeline you can open.

The setup moved this year, and a lot of guides still describe the old Developer mode switch. Here are the current steps and what to ask once you're in.

What changed in ChatGPT's MCP setup

Older guides tell you to turn on Developer mode under Apps settings. OpenAI's current docs describe a different path: custom MCP servers are added from ChatGPT's Plugins page, on the web, as a plugin. OpenAI's help center also notes that on Business, Enterprise and Edu workspaces, admins and owners control who can create and publish these connections. If you're on a company workspace and don't see the option, that's usually why.

ChatGPT only connects to remote servers, which is fine here: Rend's server is hosted.

How to add Rend to ChatGPT

These follow OpenAI's "Connect and test your plugin" steps as of November 2026:

  1. On ChatGPT on the web, open Plugins.
  2. Click the plus button and choose Add custom MCP server.
  3. Give it a name (Rend) and a short description.
  4. Under Connection, enter https://rend.ai/api/mcp.
  5. Choose OAuth for authentication, review OpenAI's risk warning, and create it.
  6. ChatGPT sends you to a Rend page. Sign in or create a free account, review the request and approve.
  7. In a new chat, call the plugin and ask for your video.

Rend supports OAuth 2.1 discovery, dynamic client registration and PKCE, so the URL is all ChatGPT needs. If a client can't do OAuth, Rend also accepts a personal token created in Studio's agent connection settings, sent as a bearer token. Keep that token in the client's credential settings, never in a prompt.

What ChatGPT can do through Rend

Rend is an AI video agent with a crew of specialist agents behind it, and connecting ChatGPT gives it storage, rendering and the whole toolset. In practice:

  • Transcribe footage with speaker labels and cut on exact words.
  • Add word-timed captions in your style, editable any time.
  • Write and voice narration in a voice you approve, designed from a description or cloned from a recording with consent.
  • Score with original music, add sound effects, and mix with ducking on separate tracks.
  • Generate shots, images and b-roll to fill gaps.
  • Research the web, X posts and Meta ads before writing hooks.
  • Export, then return a Studio link to the same editable timeline.

Generation and rendering run as background jobs. ChatGPT follows them with rend_get_job and reports back when the file is ready.

Five prompts to start with

Copy these and swap in your details.

Turn a product page into ads "Here's our product page: [link]. Make three 15-second vertical video ads with different hooks. Use our brand colors. Show me the three scripts before rendering anything."

Clean up a webinar "I uploaded the 52-minute webinar recording. Cut it to the best 8 minutes for YouTube, 16:9, 30 fps. Remove the Q&A setup and any dead air longer than two seconds. Add chapter titles."

Make a faceless explainer "Make a 4-minute explainer on how credit card interest compounds. Calm female narrator, simple motion graphics, captions on. Outline first."

Resize a winner "Take the export from the last project and make 9:16, 1:1 and 4:5 versions. Keep captions inside the safe zone."

Fix from feedback "Read the notes on the review link and make every fix. Tell me which notes you didn't act on and why."

The pattern across all five: give the length, the shape, the source and a checkpoint. A checkpoint ("show me the scripts first") costs nothing because editing on Rend is free, and it stops you paying for renders of the wrong idea. Rend charges only when it renders or generates something; see pricing.

ChatGPT with Rend vs other video connectors

Several editors now ship an MCP server ChatGPT can use. They aren't interchangeable.

ConnectorBest forWhere it's thinner
RendIdea to finished video: script, voice, music, generated shots, captions, export, review notesOverkill for a single trim
DescriptTranscript edits for teams already in DescriptWorks inside Descript's own project model
KapwingSubtitles, translation and dubbing in a Kapwing workspaceLess about building a video from a brief
OpusClipClipping long videos into shortsNarrow by design
invideoTemplate-driven edits in invideo's editorYou work in invideo's editor and plans

If all you do is subtitle and translate, Kapwing may be the simpler choice. If you cut podcasts into clips daily, look at OpusClip alternatives and podcast clips. For ad teams, the useful bit of Rend is volume: bring one video and a brand, and it spins new hooks, sizes, image ads and search ads (how to make 100 ad variations from one video).

A checklist before your first ChatGPT edit

  • Your footage is uploaded to Rend, or you have direct HTTPS links to the files (whether ChatGPT passes chat attachments to tools depends on the client).
  • You know the aspect ratio: 9:16 for Reels, Shorts and TikTok, 16:9 for YouTube, 4:5 or 1:1 for feed ads. The ad size checker shows safe zones.
  • You know the length you want, in seconds.
  • You've said what must stay in (a line, a product shot, a logo).
  • You asked for a script or outline before the render.
  • You know where you'll review: the Studio link or the review link.

Where ChatGPT video editing falls short

It's not magic. Taste still comes from you. If your prompt is "make it good", you'll get something generic. And if you need an editor's eye on pacing in a narrative film, a person will do better than any chat. ChatGPT plus a good server is strongest on work that's repeatable and clearly specified: cutdowns, captions, resizes, versions, ad variants, explainers.

There's also the plan question. What you can do with custom MCP servers in ChatGPT depends on your OpenAI plan and workspace policies, so check yours before you build a workflow around it.

For a Claude-specific version of this guide, see Claude video editing. For an overview of every way to plug an agent into video, read MCP video editing.

FAQ

Can ChatGPT edit a video I upload?

Only with a connected tool. ChatGPT alone can describe edits but can't render them. With a video MCP server connected, it can work on footage the server can reach, such as files uploaded to Rend or direct HTTPS media links.

Is Developer mode still needed for MCP in ChatGPT?

OpenAI's current docs add custom MCP servers from the Plugins page with "Add custom MCP server". Older guides referencing a Developer mode toggle are out of date.

Does ChatGPT generate the video or does Rend?

Rend does the rendering, generation and storage. ChatGPT decides what to ask for and calls Rend's tools.

Can I open the result in a normal editor?

Yes. Rend returns a Studio link to the same timeline ChatGPT built: clips, layers, audio and captions. Edit by hand and export again.

Can I use ChatGPT on my phone for this?

Adding custom MCP servers happens on ChatGPT on the web per OpenAI's docs. You can also send clips to Rend from your phone directly; see sending videos from your phone.

To try it, make a free Rend account, add the server in ChatGPT, and ask for a 15-second cut of a clip you already have. You'll know in one session whether it fits how you work.

Give Rend the job.

Editing is free. You pay only when Rend renders or generates something.