OnVid
OnVid for ChatGPT & Claude

Create AI videos right inside ChatGPT and Claude

Connect your OnVid account once, then ask your assistant for videos, animated photos, talking avatars and multi-shot ads. It picks the model, writes the prompt, shows the cost and delivers the video in the chat.

Server URLhttps://mcp.onvid.ai/mcp
How it works

From chat to finished video in three steps

  1. 1

    Connect once

    Add OnVid to Claude or ChatGPT with one URL, then sign in to your OnVid account and approve access.

  2. 2

    Ask in plain words

    Describe the video, attach a photo, or ask for a multi-shot ad. The assistant picks the model, writes the prompt and shows the cost.

  3. 3

    Watch it render

    A live card updates until the video is ready — usually 1–5 minutes. Everything also lands in your OnVid library.

Examples

What you can ask for

Real OnVid clips and the kind of request that makes them. Copy a prompt into your assistant to try it.

Product ad

Make a 5 second product shot: a sleek VR headset rotates on a pedestal, neon studio lighting.

Sample clip · Wan 2.7
Food & drink

Create a macro slow-motion espresso pour into a glass, golden crema swirling, warm backlight, black background.

Sample clip · Wan 2.7
Travel & scenery

Make a 10 second cinematic shot of an ancient cliffside temple at sunrise above a sea of clouds, slow push-in.

Sample clip · Veo 3.1
Wildlife

Generate a photoreal clip of a lion and tiger prowling in golden light, shallow focus.

Sample clip · Seedance 2
Social & UGC

Make a clip of a skateboarder landing tricks down a sunlit city street, dynamic low tracking shot, energetic.

Sample clip · Hailuo 2.3
Animation

Create a Ghibli-style scene: a child and a dog run down a lush grassy hill under big fluffy clouds.

Sample clip · Sora 2
What it can do

Your whole video studio, one message away

  • Text to video with Veo, Sora, Kling, Seedance, MiniMax, Wan, PixVerse and more — 16:9, 9:16 or 1:1, up to 1080p/4K, with native audio on supported models.
  • Animate photos and keep a character or product consistent across shots with reference images.
  • Talking avatars that speak your script in any language, with library presenters or your own portrait.
  • Multi-shot productions planned by the OnVid Director — scenes, storyboard frames and a cost estimate you approve before rendering.
  • Long videos of 30 or 60 seconds as one continuous chained clip (Max plan).
  • Free preview stills, prompt enhancement, model comparisons and cost estimates.
Compare

Claude, ChatGPT or Claude Code?

Same OnVid account, credits and library everywhere. Pick the assistant you already use.

ClaudeChatGPTClaude Code
Where it worksclaude.ai, Claude Desktop and mobile appschatgpt.com on the webYour terminal and IDE
PlansFree (1 custom connector), Pro, Max, Team, EnterprisePlus, Pro, Business, Enterprise, Edu (Developer mode)Any plan with Claude Code
How you add itCustomize → Connectors → Add custom connectorDeveloper mode → Plugins → Create appOne terminal command
ResultsVideo links in the reply, plus an interactive card where supportedInteractive card that updates until the video is readyVideo links in the reply
Your own mediaPublic image or video linksAttach images and videosPublic image or video links
Tools

14 OnVid tools your assistant can use

Read tools only look things up. Create tools make something and may use credits — each result shows its cost.

ToolTypeWhat it does
create_videoCreateMake a video from a prompt, animate a start image, keep a character or product consistent with reference images, or restyle a source video. 30- and 60-second long videos on Max.
create_avatar_videoCreateA presenter speaks your script in any language, using a library avatar or your own portrait and a chosen voice.
start_productionCreateThe OnVid Director plans a multi-shot film or ad: scenes, shots, a model per shot, storyboard frames and a credit estimate.
revise_productionCreateAnswer the director's questions or change the plan in plain words — style, length, scenes or models.
render_productionCreateRender an approved production. Shots render in sequence and are merged into one final video.
create_preview_imageCreateA free still frame or storyboard of an idea (no credits) that you can animate next.
enhance_promptReadRewrite a rough idea into a camera-ready video prompt, or notes into a spoken presenter script.
list_modelsReadVideo models with durations, resolutions, aspect ratios, audio, plan tier and price per second.
estimate_costReadPrice a video before creating it and check that your balance covers it.
list_avatarsReadPresenter avatars from the OnVid library and your saved avatars.
list_voicesReadVoices for talking avatars, including your cloned voices.
get_generationReadStatus and link of a video, preview still or production.
list_my_videosReadYour recent OnVid videos with status, prompt, model and links.
get_accountReadYour plan, subscription status and credit balance.
Privacy & safety

Your account, your credits, your call

You approve the connection

You sign in on an OnVid page (OAuth 2.0 with PKCE). The assistant never sees your password.

Costs shown up front

Every video shows its credits, multi-shot productions wait for your approval, and retries never charge twice.

Only what the tool needs

The connector receives just the inputs of each request — never your chat history.

Disconnect any time

Remove access from the OnVid app in one click; tokens stop working immediately. Failed videos are refunded automatically.

Read the privacy policy and terms of service.

FAQ

OnVid for ChatGPT & Claude — FAQ

Is OnVid an official Claude or ChatGPT app?
OnVid is an independent video platform. You add it yourself as a custom connector (Claude) or a developer-mode app (ChatGPT) using the server URL on this page. It connects only to your own OnVid account.
Does it cost extra to use OnVid inside Claude or ChatGPT?
No. The connector is free. Videos use your normal OnVid credits at the same price as in the OnVid app, and free preview stills cost nothing. Your Claude or ChatGPT plan is billed separately by Anthropic or OpenAI as usual.
Do I need an OnVid subscription?
To render videos, yes — an active Pro or Max plan. Without one you can still sign in, browse models, get cost estimates and create free preview stills.
How do I know what a video will cost before it's made?
Ask the assistant to estimate first ("how many credits would a 10 second 1080p clip cost?"). Every result also shows the exact credits used, and multi-shot productions always show the plan and estimate before anything renders.
Where do my videos go?
Straight to your OnVid library, exactly like videos made in the app. The chat shows a live card that updates until the video is ready.
How long does a video take?
Most clips take 1–5 minutes; long 30/60-second videos and multi-shot productions take longer. You can keep chatting while it renders.
Can I attach my own photos or videos?
Yes. In ChatGPT, attach an image to animate it or use it as a character/product reference, or a video to restyle it. In Claude and Claude Code, paste a public link to the image or video instead.
What happens if a generation fails?
The credits for it are refunded automatically. If the assistant retries the same request, OnVid reuses the original job, so you're never charged twice.
How do I disconnect?
Open ChatGPT & Claude in the OnVid app sidebar and click Disconnect next to the assistant — access stops immediately. You can also remove the connector in Claude or ChatGPT.

Rather not set anything up?

The OnVid Agent does the same inside the OnVid app — just sign in and chat.

Open the OnVid Agent →

Start creating from your chat

Create a free OnVid account, connect it to Claude or ChatGPT and describe your first video.

Free to sign up · Preview stills are free · Videos use your plan's credits