← All companies

World models / Company profile

Twelve Labs

What Twelve Labs does

TwelveLabs (TwelveLabs, Inc.) is a private video-intelligence company building multimodal, video-native foundation models and an enterprise “intelligence layer” for making large video libraries searchable, segmentable, and actionable. The company’s core platform combines (1) perception-style models that convert video into embeddings and structured representations and (2) reasoning workflows that let users query across an indexed video corpus using natural language. TwelveLabs positions its approach as video-first multimodality—i.e., models designed to understand motion, audio, and on-screen text in the context of the video timeline—rather than language models that only “look at” sampled frames. TwelveLabs’ products are exposed through REST APIs and SDKs via its developer documentation, with additional application-layer functionality for creative workflows.

From a product standpoint, TwelveLabs offers two main interaction styles: “Models” for dedicated tasks (search across moments, analyze individual videos, and generate embeddings) and “Agents” (a research-preview unified system, Jockey) designed to reason over an indexed knowledge store across multiple modalities and return grounded, cited moments. The company’s named foundation models include Marengo (an embedding model for comprehensive video understanding) and Pegasus (a generative video-to-text model). In addition to the model layer and API access, TwelveLabs announced Rodeo as its first application-layer product for creators—an AI rough-cut assistant that turns plain-language story direction into an assembled rough cut, followed by conversational refinement and export.

TwelveLabs primarily targets developers and enterprise teams managing large-scale video workflows—media & entertainment, advertising, sports & broadcasting, security, and government are explicitly cited across its platform/press materials. Business model details are not fully disclosed in the sources captured here, but the company markets the platform via “Talk to Sales” for enterprise deployment and provides API/SDK access for developers, suggesting consumption-based usage (API plans) alongside enterprise contracts for deployment where customer data lives.

News

Company record

Jul 01, 2026 · Official · GlobeNewswire (Twelve Labs, Inc.)TwelveLabs Raises $100 Million in Series B Funding to Build Video Superintelligence

TwelveLabs announced a $100M Series B co-led by NEA and NAVER Ventures, with participation from Amazon, Radical Ventures, Korea Investment Partners, Index Ventures, Quadrille Capital, and Red Bull Ventures. The company described expanding beyond models into a full-stack agentic intelligence system for video.

Jun 01, 2026 · Official · PRWeb (News provided by TwelveLabs)TwelveLabs Bring Its Video Understanding Technology Directly to Creators (Rodeo availability)

TwelveLabs announced the availability of Rodeo, described as its first application-layer product—an AI rough cut assistant for creators that assembles clips into story-driven rough cuts using natural-language direction.

Dec 12, 2024 · Official · TwelveLabs Blog (PRNewswire-PRWeb republish content)TwelveLabs Secures $30M in Funding to Advance Video AI (Strategic investments + Yoon Kim hire)

TwelveLabs announced $30M in strategic investments from Databricks, Snowflake, SK Telecom, HubSpot Ventures, and IQT and hired Yoon Kim as President and Chief Strategy Officer.

Mar 16, 2022 · Official · TwelveLabs BlogTo make the world’s videos searchable, Twelve Labs raises $5M seed round led by Index Ventures

TwelveLabs announced its $5M seed round led by Index Ventures to build video understanding infrastructure for developers.

Date not disclosed · Official · TwelveLabs BlogTwelveLabs lands $12M for AI that understands the context of videos (seed extension led by Radical Ventures)

TwelveLabs announced a $12M seed extension led by Radical Ventures aimed at extracting movement, objects, sound, and speech to enable semantic search and video applications like summarization, chapterization, and question answering.

Source map · 3 recurring channels · 19 references

Still resolving: Newsroom

Funding

Latest disclosed valuationNot disclosed
Tracked capital$197M6 sourced rounds

What it builds

TwelveLabs video intelligence platform (API/SDK)

A video intelligence platform that supports uploading video libraries and then using models or agents to search, analyze, and generate embeddings via REST APIs and Python/Node.js SDKs.

source ↗
Marengo (embedding model, current version Marengo 3.0)

A multimodal embedding model for comprehensive video understanding, designed to support fine-grained search (including logos/text/small objects), motion search, object counting, and audio comprehension, with long-content support.

source ↗
Pegasus (generative video-to-text model, current version Pegasus 1.5)

A generative model for video-to-text generation that analyzes multiple modalities to produce contextually relevant text outputs, including scene/content descriptions and timestamp-grounded answers.

source ↗
Jockey (agentic system, research preview)

A unified agentic system that reasons across a user’s videos and images by querying a knowledge store with natural-language instructions and returning structured JSON outputs when a JSON schema is provided.

source ↗
Rodeo (AI rough cut assistant)

An application-layer “rough cut” assistant where users describe the story in plain language; Rodeo finds matching clips, assembles them into scenes, supports conversational/timeline refinement, and exports to finishing tools.

source ↗

Milestones & partnerships