Meta Muse API examples (useapi.net): AI video, images and agent chat on muse.ai

Runnable examples for the Muse API by useapi.net: make images and 10-second videos with sound using Muse (muse.ai), Meta's AI agent, through a REST API that runs on your Muse subscription: Free, Power $20 or Maximum $100 a month (as of October 2026). Combine up to 9 photos into one picture, restore and colorize old photos, animate a still with spoken dialogue, and replace a person in a video. Every file Muse makes is an asset id the next job takes as an input, so edits chain with no downloads or re-uploads.

Demo on YouTube: Meta Muse API Demo: Images, Talking Photos, Video with Sound, Chat & MCP (third-party, useapi.net)

| Example | What it does | Tutorial | Tutorial date | |---|---|---|---| | images-videos/ | Three photos become one picture of a rider on a vintage motorcycle in the surf, the picture becomes a 10-second video in which she shouts a line, and a woman from a fourth photo replaces her in that video. Also two images from text, and a hundred-year-old portrait restored, colorized and brought to life. Node.js and Python | How to Generate AI Videos and Images with the Meta Muse API | October 12, 2026 | | narrated-short/ | Narrated 30-second Short from a topic: a chat writes the script as JSON, three 9:16 video jobs run at once with the same off-screen narrator speaking each line, and a last chat joins the clips with ffmpeg into one MP4. About 2 to 3 minutes per Short. Optional fixed voice through Muse text to speech. Node.js and Python | Make a 30-second narrated Short | October 18, 2026 | | demo/ | Live chat app: one self-contained HTML page that chats with the Muse agent through the API, with answers streamed live, files and cards, approvals, several live chats at once, workspace files and the agent's permissions. Open the live demo or open demo/index.html locally. Vanilla JS, no build | Chat Streaming Guide | October 12, 2026 | | skills/muse-video/ | Agent skill for Claude Code, Codex and other coding agents: ask for an image or a clip and the agent makes it with Muse (text to image, photo compositing, old-photo restoration, talking photos, person swap in a video) and downloads the .webp or .mp4, never paying twice for a request. Install: npx skills add useapi/muse-api --skill muse-video. Bash, curl, jq | SKILL.md | October 12, 2026 | | n8n/ | n8n app: upload an old photo, Muse restores and colorizes it, you pick or redo, then the person in it says your line in a 10-second video with sound. 1 or 2 takes, the real share of the week used on the result page, optional Google Drive save, and a Google Sheet batch mode. Core n8n nodes, no community nodes. Plus narrated faceless Shorts from a topic: a script page, one narrated 9:16 scene per part, joined and saved to Google Drive | n8n template #20989, #21207 | October 19, 2026 | | MCP server | Hosted MCP server: connect Claude Code, Cursor, VS Code or Windsurf to https://api.useapi.net/v1/muse/mcp and the assistant gets 37 tools for Muse images, videos, chat, files and permissions. Nothing to install or host | MCP Server | October 13, 2026 |

Quick start

You need Node.js v21 or newer or Python 3.x (no dependencies to install), a useapi.net API token, and a Muse account (free or paid) added with the Muse setup.

git clone https://github.com/useapi/muse-api.git
cd muse-api/images-videos
node ./images-videos.mjs <API_TOKEN> <EMAIL>
# or, equivalently, with Python:
python3 ./images-videos.py <API_TOKEN> <EMAIL>
# just one image and one video (restore an old photo, then make it talk):
node ./images-videos.mjs <API_TOKEN> <EMAIL> --steps portrait-video

Each example's README lists its options. Every endpoint is documented in the Muse API reference.

MCP server

useapi.net also runs a hosted Model Context Protocol server for Muse, so an AI assistant such as Claude Code, Cursor or VS Code Copilot can use Muse on your Muse subscription. It has 37 tools: generate images and videos, chat with the Muse agent, answer its permission requests, open a live view of its web browser, manage the account's Library and files, and reset the account. Each tool calls the Muse API, so the same rules, limits and allowance apply. Nothing needs to be installed or hosted.

• Endpoint: https://api.useapi.net/v1/muse/mcp
• Transport: Streamable HTTP
• Authorization: your useapi.net API token in the Authorization: Bearer … header, and at least one Muse account added with the Muse setup

Claude Code:

claude mcp add --transport http useapi-muse https://api.useapi.net/v1/muse/mcp --header "Authorization: Bearer YOUR_API_TOKEN"

Cursor (~/.cursor/mcp.json):

{
  "mcpServers": {
    "useapi-muse": {
      "url": "https://api.useapi.net/v1/muse/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_TOKEN"
      }
    }
  }
}

VS Code (.vscode/mcp.json):

{
  "servers": {
    "useapi-muse": {
      "type": "http",
      "url": "https://api.useapi.net/v1/muse/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_TOKEN"
      }
    }
  }
}

Windsurf setup, a curl test, the full list of 37 tools and how long-running videos and chats are handled: Muse MCP Server.

Use with Claude Code

This repo is also a Claude Code plugin marketplace. The plugin adds the hosted Muse MCP server and the muse-video agent skill, and reads your token from the USEAPI_TOKEN environment variable, so set it before you start Claude Code:

export USEAPI_TOKEN=user:12345-...      # your useapi.net API token

Then, inside Claude Code:

/plugin marketplace add useapi/muse-api
/plugin install muse@useapi-muse

On Claude Code 2.1.275 or later, /plugin install muse --marketplace useapi/muse-api does both in one step. Run /mcp to check that useapi-muse is connected.

Use with Gemini CLI

This repo is also a Gemini CLI extension with the same MCP server and the muse-video agent skill:

gemini extensions install https://github.com/useapi/muse-api

The install asks for your useapi.net API token and keeps it in the system keychain. Change it later with gemini extensions config useapi-muse, and check the connection with gemini mcp list. The muse-video skill's scripts read USEAPI_TOKEN from your shell, so export it as well if you use the skill.

Articles and videos

• Is There a Meta Muse Video API? Official API, Free Tools and a Hosted API (2026)
• How to Generate AI Videos and Images with the Meta Muse API
• Chinese-language tutorial (中文教程): AI images and videos with dialogue
• Japanese-language tutorial (日本語チュートリアル): AI images and videos with dialogue, including Japanese speech
• YouTube: Meta Muse API demo, a narrated 30-second Short made only with Muse API calls, 3 photos in, 1 talking video out, The Great Pyramid was once bright white and Why honey never spoils (both made by the n8n Faceless Shorts workflow)

Common questions

<!-- Official API figures in these answers, re-read 2026-10-20 (useapi.net/docs/api-muse-v1 sources): https://dev.meta.ai/docs/pricing-rate-limits (Muse Image $0.01 per image, no video model, no API for the Muse agent). Plans: https://www.meta.com/help/subscriptions/1021145227643680/ (Free, Power $20, Maximum $100 a month). Allowance figures: our tests on the free plan, 2026-10-12. -->

• Is there a Meta Muse API? Yes, through useapi.net: the useapi.net Muse API makes 10-second videos with sound and spoken dialogue, images, photo edits, person swaps and live chat with the Muse agent, on your Muse subscription. Meta sells no Muse video API: its Model API has the Muse Image model at $0.01 per image, but no video model and no API for the Muse agent (as of October 2026). Is There a Meta Muse Video API? compares the options.
• How much does it cost? Through useapi.net, an image or a video has no price of its own: it uses the weekly allowance of your Muse subscription, and the free plan works. In our tests on the free plan, a video used 2 to 3 percent of the week and an image 0 to 1 percent, and a paid plan gets a bigger allowance (usage limits). Meta's Model API sells no Muse video at any price (as of October 2026).
• Is there a Muse API key? Through useapi.net, one useapi.net API token is all the key you need, for the REST API and the MCP server alike. Add your muse.ai email and password once with Setup Muse, and we keep the account signed in, with no cookies to copy. Create the account in the US, Canada, Japan or Singapore, where Muse is available (as of October 2026), then call the API from anywhere.
• How do I set the aspect ratio, length or resolution? Through useapi.net, in the prom