The Best Video MCP Servers in 2026, Sorted by What You Get Back

November 16, 2026

A video MCP server lets an AI assistant like Claude or ChatGPT make or edit videos for you while you chat. You connect it once, sign in, and from then on "make a 30-second ad for my product" turns into an actual video instead of a script you still have to film.

The catch is that "video MCP server" covers very different things. Some hand back a finished video with a voiceover, captions and music. Others hand back a five-second AI clip, a presenter reading your script, or a render of a template you designed first. This guide sorts nearly 40 of them by what your assistant actually gets back, so you can pick the right kind before comparing brands.

We make Cliptalk, one of the servers in the first group. Everything below was checked against each company's own pages on November 16, 2026: setup docs, help centers, pricing pages and their hosted server addresses. We didn't sign up for paid plans or run paid jobs, so where a company doesn't say something, we say so. Plans and prices change often; check the vendor's page before you buy.

The short answer

What you wantStart with
A whole finished video from one messageCliptalk, Revid, MakeAIVideo, Zebracat, Pictory
A presenter reading your scriptHeyGen, Synthesia, VEED Fabric
One striking AI shot or clipRunway, Luma, Kling, Pika
Many AI models in one placefal, Pika, Higgsfield, Replicate
The same branded template, many timesShotstack, Creatomate, JSON2Video
Edit footage you already shotDescript, Kapwing, invideo
Short clips cut from a long video, posted for youOpusClip, Reap

1. Finished-video makers

You describe the video, and the server writes the script (or takes yours), makes the scenes, adds a voiceover, captions and music, and hands back an MP4. This is the group to look at if you want to post what comes back.

ServerWrites the scriptChange it afterPosts to social appsFree to try through the server
CliptalkYes, or keeps yours word for wordBy chat or in its editor, with undoNoYes: free credits, no card, videos up to 60 seconds
RevidYesBy chatYes, now or scheduledNo: making videos needs a paid plan
MakeAIVideoYesBy chatYes, now or scheduled7-day trial, card required
ZebracatYesIn its web editorNot through the serverNot stated
PictoryNot stated who drafts itPreviews by chat, then in its editorNot mentionedYes, on its free trial
Agent OpusYesIn the Agent Opus appYesIts page says Pro plans and up
SynthesysYes, from one briefIn its web appNoNot stated
Lumen5Yes, from your text or notesIn Lumen5NoNot stated
AITuberYesNot documentedYouTube, TikTok, InstagramYes: 100 free credits, no card
Faceless.soTakes a finished scriptIn its Studio editorYes, seven networksNo: the free plan has no credits
AutoShortYesOne scene at a time, by chatYouTube; TikTok as draftsNo free plan
VideoGenNarrates the script you give itBy chatNoNot stated
AIVideo.dayYesNo edit toolNoPaid plans only

Cliptalk (that's us) makes short videos for TikTok, Reels and Shorts from an idea, a script, a web page or your own pictures. Your assistant can ask for changes afterwards ("shorter hook", "swap the music", "redo the third clip"), and earlier versions are kept, so a change can be undone. It connects to Claude, ChatGPT, Grok, Cursor, VS Code, Claude Code, Codex and Gemini CLI, with a sign-in instead of an API key. It doesn't post to social apps: you download the MP4 and post it yourself.

Revid is the closest match to Cliptalk and goes further on distribution: it can research viral videos in your niche first, and post or schedule the result to TikTok, YouTube, Instagram, X and Facebook Pages. Making videos through the server needs an active paid plan.

MakeAIVideo covers explainers, listicles, stories, UGC and spokesperson videos, can regenerate a single scene or voice, and publishes to more than ten networks. Its trial asks for a card.

Zebracat writes, voices and captions the video and opens it in its web editor for changes. It's hosted in Germany.

Pictory returns an MP4 and a shareable player link, and you can change a preview by chat before saving it. Which kind of video you get depends on your account version: Pictory 3.0 accounts make motion-graphics videos only.

Agent Opus, from the makers of OpusClip, turns a topic or script into a finished video and posts to seven networks. Edits happen in its own app rather than through the server.

The rest of this group: Synthesys quotes a price before it builds and can remake an ad from a link. Lumen5 turns articles and notes into branded business videos inside your Lumen5 workspace. AITuber and Faceless.so focus on faceless shorts and post them for you. AutoShort makes 9:16 Shorts and uploads to YouTube. VideoGen narrates the script your assistant writes and has a large media toolkit. AIVideo.day does long narrated videos and 3D films, rendered in a browser page.

2. Avatar and presenter tools

These hand back a video of an AI presenter speaking. Two of them now also have agents that build a multi-scene video around the presenter.

ServerWhat you get backChange it afterFree to try through the server
HeyGenAvatar videos from a script; its Video Agent makes multi-scene videos from a promptVideo Agent videos, scene by scene by chatYes: 3 watermarked videos a month, up to 1 minute
SynthesiaAvatar drafts from templates; its Syren agent makes videos from a briefSyren videos by chat, with versions; drafts in its editorDrafts need a paid plan; Syren uses free promo credits first
VEED FabricA talking head of a stock character reading your scriptNo edit toolNeeds AI Playground credits

HeyGen has the widest toolkit here, with over 100 tools: avatar videos, translation with lip-sync, clipping long videos, voice cloning and templates. Synthesia says Claude and ChatGPT are the only apps it tests against. VEED Fabric is the simplest: pick a character and a voice, give it a script, get an MP4. It doesn't add captions or music.

3. Clip generators and model hubs

These make the raw footage: a few seconds of AI video from a prompt or a picture, often with sound. Your assistant writes each shot. To turn the clips into a finished video you still need an editor, or one of the servers in the first group.

ServerWhat it's good atSign-in
RunwayImages and video from Runway's models and others, saved to your Runway libraryAccount sign-in
LumaClips from a description or image; extend, restyle or reframe themAccount sign-in
KlingText-to-video and image-to-video with Kling's modelsAccount sign-in
Pika60+ apps and models, multi-shot clips up to 30 seconds, speech and musicAccount sign-in
LTXGenerate clips, then extend, retake, reframe, relight or add sound to themAccount sign-in
HiggsfieldIts model lineup plus ad presets from a product linkAccount sign-in; paid plan required
fal1,000+ models, with a price you can check before each runSign-in or API key
ReplicateThousands of models through Replicate's APIAPI token
HedraImages, video, speech, music and sound effectsAccount sign-in
ArgilIts playground's models, including talking avatarsSign-in or API key; active plan to generate
ElevenLabsSpeech, sound and music, plus image and video clips on an editable canvasAccount sign-in
MiniMaxHailuo video, speech and voice cloningRuns on your computer with an API key

Runway launched its server in May 2026, and every generation through it uses credits, even on plans with unlimited use in the app. Luma works on every Luma plan and can edit and extend what it makes. Pika is the most "all-in-one" of the clip tools: its Product Ad app starts from a product image and a brief, and in ChatGPT it can bring up a timeline. fal suits people who want to choose between many models. One thing to know about ElevenLabs: the version listed in Claude's connector directory can't make images, video or music; you need its custom connector address for those.

4. Template renderers

You (or your assistant) design a template or a timeline once, then the server fills in new text, pictures and voice and renders a video. Great for the same branded video many times over; your assistant writes every line and timing.

ServerVoice, captions and musicFree to try
ShotstackAI voice, auto captions and AI music as timeline layers10 free credits for 30 days; free sandbox renders are watermarked
CreatomateAuto subtitles; AI voice uses your own ElevenLabs or OpenAI key50 free credits; free renders are small previews
JSON2VideoAI voiceover and auto subtitles600 free credits, watermarked
PlainlyNot statedFree on any plan, including the trial

Shotstack can show an editing canvas right inside the chat; its server is in beta. Creatomate and JSON2Video are hosted with a sign-in. Plainly renders Adobe After Effects templates and runs on your own computer with an API key. Canva also has a server, and it can export a design as MP4, but it doesn't generate video.

5. Footage editors and clippers

These start from video you already have.

  • Descript imports, transcribes and edits footage with its Underlord assistant, adds captions and publishes to a Descript link. The version in Claude's directory leaves out its generative image and video tools.
  • Kapwing subtitles, translates, dubs and edits videos in your Kapwing workspace, and has added AI generation. Its MCP page says it works on all plans, with export limits on free accounts (an older help article still says Pro and up).
  • invideo lets your assistant edit a live timeline while you watch, then export an MP4 or a Premiere, Final Cut or DaVinci Resolve project. Your assistant writes the script, and invideo never posts.
  • OpusClip and Reap cut short clips from a long video, caption them and post or schedule them. Reap's server works on its free plan; OpusClip's tools need a Pro or higher plan.
  • Clueso makes product walkthroughs and tutorials for software companies, and Adobe's connector crops and resizes clips for Shorts and Reels. Adobe has announced servers for Premiere and After Effects, in private beta from early 2027.

What isn't on the list

  • OpenAI Sora: OpenAI shut down its video API on September 24, 2026, and there was never an official Sora server.
  • Google Veo: Google's routes are for developers on Google Cloud, and its open-source video servers are marked "not an officially supported Google product".
  • CapCut has a ChatGPT and Codex plugin, not an open server you can add to any app.
  • Tavus builds live video agents rather than video files, and D-ID's server is for developers working with its API.
  • We found no official server for Colossyan, Elai, Clipchamp, Fliki, Visla, Vidnoz, Steve.ai or Vadoo, and Remotion shut its documentation server down in October 2026.

How to choose

  1. Decide what you need back. If you'll post what comes out, start with the finished-video makers. If you're cutting your own edit, a clip generator gives you better raw shots.
  2. Check that it works in your app. Most support Claude and ChatGPT; fewer document Grok, Gemini CLI or VS Code.
  3. See whether you can try it free through the server, not just in the website. Several need a paid plan before the assistant can make anything.
  4. Ask for the price first. Many of these can quote a job before running it; ask your assistant to do that every time.
  5. Prefer a sign-in over an API key if you're not a developer. It's simpler, and there's no secret to copy around.

Connect one in a minute

For Cliptalk, open cliptalk.pro/mcp, pick your app and click the button. Claude opens with everything filled in; Grok and ChatGPT open with the message typed for you. Then ask for your first video in plain words. New accounts start with free credits, so you can see a finished video before you pay for anything.

More articles

Join 4500+ creators