> This page is for version v1.3 (default).
> For other versions, use one of these documentation indexes:
> - v1.3 (default): https://beta.docs.twelvelabs.io/v1.3/llms.txt

> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://beta.docs.twelvelabs.io/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://beta.docs.twelvelabs.io/_mcp/server.

# Choose between Models and Agents

> Choose Models for search, analysis, segmentation, and embeddings, or Agents (Jockey) for reasoning across your videos and images.

TwelveLabs offers two ways to work with your content. Models work on one video at a time. Even when you search an index of many videos, each result comes from a single video, and you combine the results yourself. Agents (Jockey) work across all the videos and images in a knowledge store and can reason across them.

#### [Models](/v1.3/docs/get-started/quickstart)

Dedicated APIs for search, analysis, segmentation, and embeddings. You call each one and combine the results yourself.

#### [Agents (Jockey)](/v1.3/agents/get-started/quickstart)

A unified agentic system that reasons across your videos and images. Ask a question, and Jockey plans its own steps, then answers with grounded, cited moments.

# Which one fits your task?

**Models**

* Search for moments across your videos with text or image queries
* Analyze a single video or one or more images for summaries or answers to your prompts
* Segment a video into labeled, timestamped parts
* Generate vector embeddings for your machine-learning pipeline

Start with the [Playground](/v1.3/docs/resources/playground) or the [Models API](/v1.3/docs/get-started/quickstart).

**Agents**

* Have [Jockey](/v1.3/agents/concepts/jockey) reason over a knowledge store
* Ask a question that spans every video and image in the store
* Reason through a complex question that no single search or analysis can answer
* Track a person or object across multiple videos
* Organize or classify a video library by topic
* Build multi-step workflows that return structured JSON output

Start with the [Jockey MCP](/v1.3/agents/get-started/start-with-jockey) or the [Agents API](/v1.3/agents/get-started/quickstart).

**Need both?** Use Models for embeddings and per-video analysis, and Agents for corpus-level reasoning. Both start from an asset, a file you upload once, so you can use the same asset in each.

Already using Models? See [Migrate from Models](/v1.3/agents/get-started/migrate-from-models) to compare the two and move your workflows.