Luma API MCP
About
Powered by Ray (video) and Photon (image) models by Luma AI
Details
- Author
- lumalabs
- GitHub stars
- 25
- Downloads
- 571
- Categories
- Developer Tools, API
Jump to
- Create images with custom prompts and aspect ratios
- Create videos with configurable duration, resolution, and looping
- Supports multiple models: photon‑1/flash for images, ray‑2/flash/1‑6 for videos
- Reference up to eight images, one style, four character images per generation
- Video keyframe control using images or prior generation IDs
- Typical generation times: images 5–15 s, videos 15–60 s
Install by running setup.sh and pasting your API key (obtain it from https://lumalabs.ai/api/keys). Then open any MCP client, such as Claude Desktop, and invoke the “Create Image” or “Create Video” tools with the parameters described in the README.
Luma API MCP
Setup
1. Install Claude Desktop App or any MCP client
2. Get API Key from https://lumalabs.ai/api/keys
3. Run sh setup.sh, here it will ask the API Key - paste it from 2.
Run
Open Claude Desktop or any MCP client
Features
Create Image
- prompt: text
- aspect_ratio: "1:1", "16:9", "9:16", "4:3", "3:4", "21:9", "9:21" (default: "16:9")
- model: "photon-1", "photon-flash-1" (default: "photon-1")
- image_ref: list of image URLs with weights to influence generation (optional), max 8
- style_ref: single image URL with weight to influence style (optional), max 1
- character_ref: list of character image URLs (optional), max 4
- modify_image_ref: single image URL to modify with weight (optional), max 1
Create Video
- prompt: text
- aspect_ratio: "1:1", "16:9", "9:16", "4:3", "3:4", "21:9", "9:21" (default: "16:9")
- model: "ray-2", "ray-flash-2", "ray-1-6" (default: "ray-2")
- loop: boolean (default: false)
- resolution: "540p", "720p", "1080p", "4k" (default: "720p")
- duration: "5s", "9s" (default: "5s")
- frame0_image: image URL (optional)
- frame1_image: image URL (optional)
- frame0_id: generation ID (optional)
- frame1_id: generation ID (optional)
Technical Notes
- Keyframes: Providing frame0_image/frame1_image gives more control over video start/end points - Video Response Time: Typical video generation takes 15-60 seconds depending on duration and resolution - Image Response Time: Typical generation takes 5-15 seconds depending on model complexitySign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.





