TwelveLabs documentation
Search, understand, and build with video
TwelveLabs finds the exact moments in your videos and answers questions about what happens in it. Use Jockey from Claude, ChatGPT, or the Agents API. Or build with the Marengo and Pegasus models through the Models API.
Jockey or Models?
Every path below follows the same four steps. Choose Jockey or Models, then choose how you want to use it. Not sure which to choose? See Choose between Models and Agents.
- 1Set upAdd Jockey to Claude or ChatGPT, sign in, or create an API key
- 2Add videosUpload your videos
- 3Ask or runAsk a question or run a task
- 4Read resultsCited moments, clips, text, or embeddings
Jockey
Ask questions across your videos and receive cited moments. Jockey plans its own steps.
Jockey MCP →
Connect Jockey to Claude or ChatGPT, then ask about your videos in plain English.
Agents API →
Create a knowledge store, then create a response or search it from Python or Node.js.
Models
Call Marengo and Pegasus directly to search, analyze, segment, or create embeddings. You choose exactly what runs.
Playground →
Try search, analysis, and segmentation on your own videos in the browser.
Models API →
Search, analyze, segment, or embed from Python or Node.js.
Jockey
A short explanation of how Jockey reasons across your videos and images.
Create a knowledge store and ask Jockey questions from your own code.
Ready-made tasks for your videos, like highlight reels, agentic search, and organizing a video library.
Check your videos in Connect
Claude and ChatGPT only show the chat. Connect shows what Jockey has stored: your knowledge stores, uploads, collections, usage, and billing. Sign in with the account you used to connect Jockey, then open a knowledge store to see your videos and usage.
Models
Search, analyze, segment, or create embeddings. Choose a task and run it.
Try search, analysis, and segmentation on your own videos in the browser before writing code.
See what your plan includes before you build.
Reference
Typed clients with full reference for every endpoint.
Endpoints, authentication, and request and response schemas.
Learn from real examples
No-code guide. Search your videos from the assistant you already use.
Developer tutorial. Build an app on Pegasus and Jockey that turns security camera video into searchable insights.
How AI-first pipelines can reduce annotation timelines from months to days.
Working projects you can clone and adapt.