Twelve Labs

What Twelve Labs does

Twelve Labs is a video intelligence platform built on three proprietary multimodal foundation models: Marengo (video understanding and search), Pegasus (video-text embeddings), and Rodeo (generative creative workflows). The models enable natural-language search, scene-level temporal localization, video classification, summarization, and metadata extraction across multi-hour videos in 36 languages, and are exposed through APIs as well as through integrations with AWS Bedrock, AWS Marketplace, NVIDIA NIM, Databricks, Snowflake, and VAST Data. Technical positioning is anchored in claimed benchmark leadership on Video-MME and a reported +13.1% advantage over Gemini 3.1 Pro on Multimodal Prompting, with production throughput of 10k+ hours of video indexed per day.

The company serves enterprise customers across media and entertainment, sports, public sector (including surveillance and defense-adjacent workloads), telecommunications, and technology. Named customers include Firstman Studio, UNICEF Korea, Quickplay, Backblaze, LG CNS, and SK Telecom, with publicly cited coverage of workflows such as sports highlight generation, content moderation, media asset management, and public-safety video review. A recent VAST Data partnership supports customer-managed deployment for buyers with data-residency requirements, and an Ecosystem Partner Program (April 2026) targets SI/ISV channels for indirect distribution.

Monetization is a usage-based/API model with enterprise contracts, distributed both direct and via cloud-marketplace channels. The company has raised over $77M to date from investors including NEA, NVentures, Databricks, Snowflake, SK Telecom, NVIDIA, Intel, Samsung, and Firstman Studio, reaching unicorn status in 2025. It is headquartered in San Francisco, was founded in 2020, and reports a headcount in the 11-50 range.

Name

Twelve Labs

Legal name

Twelve Labs, Inc.

Website

https://twelvelabs.io

Company type

Private

Founded year

2020

Operating status

Operating

Headcount range

11–50 employees

Short description

Twelve Labs is a video intelligence platform that uses proprietary multimodal foundation models (Marengo, Pegasus, Rodeo) to enable natural-language search, scene-level discovery, and metadata extraction across enterprise video libraries. It serves media, sports, public sector, and Fortune 500 customers globally.

Ownership category

akta.pro rank

Product category

Video Intelligence Platform

NAICS

Software Publishers (513210), Custom Computer Programming Services (541511)

SIC

Services-Computer Programming Services (7371)

akta.pro primary industry

Asset Creation, Pipeline, DCC Integration & Content Management Tools (MPAFAFAL)

akta.pro secondary industries

Virtual Production Production Management & Collaboration Tools (MPAHADAK), Model Development & Training Platforms (AutoML, Notebooks, Feature Stores) (HDAEANAB)

Keywords

• Video intelligence platform
• Multimodal video AI
• Video search APIs
• Video understanding models
• Video analytics software

Headquarters

HQ city

San Francisco

HQ country

United States

HQ region

North America

Offices2 records

Markets served

GTM type

B2B

Offering type

Software

Revenue model

1. API and Platform Subscription: Recurring revenue from enterprise customers accessing video understanding APIs and infrastructure for video indexing, search, and analysis. Pricing appears to be quote-based for enterprise deployments.
2. AWS Marketplace / Bedrock Distribution: Usage-based revenue from consumption of Marengo, Pegasus, and Rodeo models via Amazon Bedrock and AWS Marketplace, where AWS serves as a cloud service provider distributing the models.
3. Enterprise Licensing / Custom Deployments: Revenue from enterprise agreements including custom on-premises deployments (via VAST AI OS) for media, financial services, and public sector customers with data sovereignty requirements.

Pricing tiers

Go-to-market motion3 records

Distribution channels5 records

Marketing channels9 records

Core offering

Twelve Labs develops and sells a video intelligence platform and developer APIs built on proprietary multimodal video-language foundation models (Marengo and Pegasus) that ingest raw video and produce embeddings, semantic search results, scene-level segments, compliance flags, highlights, and timestamped structured metadata. Customers access the models via direct enterprise agreements, a self-serve Playground, REST APIs and SDKs, AWS Marketplace, and Amazon Bedrock. The company also offers Rodeo, an application-layer creative copilot for video production workflows.

Product overview

TwelveLabs offers a unified video intelligence platform built on proprietary multimodal foundation models, combining a core infrastructure layer with model APIs, an MCP server, an application-layer product, and an ecosystem partner program. The core engine is the Marengo video understanding model (multi-vector embedding, currently at v3.0) and the Pegasus video-language model (currently at v1.5, with Time Based Metadata Extraction). These power developer-facing APIs for search, embed, and generate, exposed via REST, SDKs, and a Playground, and distributed through Amazon Bedrock and AWS Marketplace. Rodeo is the first application-layer product built on top of this stack — a creative copilot for video production. The TwelveLabs Ecosystem Partner Program extends the platform by enabling SIs and ISVs to resell and build solutions on top of it.

Differentiator

Problem solved

Functional benefit

Products and services

• TwelveLabs Video Intelligence Platform Core infrastructure platform that ingests raw video, indexes it via multimodal foundation models, and exposes it through APIs, SDKs, and an MCP server for downstream search, segmentation, compliance scanning, highlight creation, and insight generation. Used by enterprises processing millions of hours of video across media, sports, and public-sector deployments.
• Marengo TwelveLabs' flagship multimodal video foundation model using a multi-vector approach to natively understand video, audio, text, and context. Supports up to four-hour videos across 36 languages; claimed #1 ranking on Video-MME benchmark. Sold via TwelveLabs API, AWS Marketplace, and Amazon Bedrock.
• Pegasus TwelveLabs' video-language foundation model for video-to-text generation and reasoning. Pegasus 1.5 introduces Time Based Metadata Extraction (TBM), producing timestamped, structured metadata from up to two hours of video in a single API call. Available via TwelveLabs API, AWS Marketplace, and Amazon Bedrock.
• Rodeo TwelveLabs' first application-layer AI-powered creative copilot for video production workflows. Lets creators find, edit, and assemble footage using natural-language prompts (e.g., producing rough cuts from 200 hours of dailies, scored-goal compilations, thematic clips) and exports directly into editing workflows.
• TwelveLabs APIs & SDKs (Search, Embed, Generate) Developer-facing suite of REST APIs and SDKs for semantic video search, embedding generation, and video-to-text generation, backed by a Playground for evaluation and an authenticated Developer Hub for production access.

Quantifiable outcome

• 95% reduction in search time for UNICEF Korea's 8TB media archive
• +4 more outcomes

Named customers4 records

Segments6 records

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration11 records

AI capability12 records

Feature6 records

Partnerships

Twelve partnerships are on record, tiered core, flagship and minor.

• FriendliAIcoreTechnology or Integration · 12 May 2026FriendliAI lists TwelveLabs among its enterprise customers, providing frontier AI inference for open-weight and custom models across AWS, OCI, and Samsung Cloud Platform.
• Sports Video Group (SVG)coreStrategic or Co-development Partner · 14 April 2026TwelveLabs is a founding technology partner of the SVG AI Innovation Lab alongside AWS, AKTA, and CoreWeave, coordinated with senior leaders from the NBA, PGA Tour, MLB, and Fox Sports to advance AI in live sports production.
• QuickplaycoreChannel Partner/ Reseller/ Distributor · 2 April 2026Quickplay joined the TwelveLabs Ecosystem Partner Program (Advanced tier) and AWS Business Outcomes Xcelerator (BOX) Program to deliver AI-powered video intelligence solutions to Tier-1 media, sports, and entertainment companies.
• VAST DataflagshipStrategic or Co-development Partner · 2 March 2026TwelveLabs and VAST Data announced a strategic partnership at VAST Forward 2026 to enable customer-managed deployment of TwelveLabs' video intelligence models on the VAST AI Operating System, which manages unstructured data at exabyte scale. The collaboration supports on-premises and neocloud deployments for media, financial services, and public sector organizations with data sovereignty requirements.
• NVIDIA (technology)flagshipTechnology or Integration · 26 February 2026VAST Data deepened NVIDIA GPU server integration through the CNode-X configuration in February 2026, while TwelveLabs participates in NVIDIA's Holoscan for Media IBC showcase. TwelveLabs models also run on NVIDIA GPUs in AWS.
• BackblazeminorGTM or Marketing Partner · 10 August 2025Backblaze co-hosted a partnership webinar showcasing AI-powered media archive solutions built with TwelveLabs video AI.
• Amazon Web Services (AWS)flagshipTechnology or Integration · 15 July 2025TwelveLabs' Marengo and Pegasus foundation models are deployed on AWS infrastructure (Amazon EKS, EC2 with NVIDIA GPUs, Amazon S3 with S3 Vectors) and made available in Amazon Bedrock as well as AWS Marketplace. AWS is the first cloud service provider to distribute the model, allowing TwelveLabs to serve enterprises processing millions of hours of video.
• LG CNScoreStrategic or Co-development Partner · 2 July 2025TwelveLabs and LG CNS partnered in July 2025 to accelerate innovation in video AI technology, advancing video understanding capabilities.
• PentagramminorOthers · 5 June 2025Pentagram designed a new brand identity for TwelveLabs in 2025, including the 'galloping horse' logo inspired by early motion studies, covered by Creative Boom, Design Week, and Creative Review.
• WizelineminorGTM or Marketing PartnerWizeline co-presented a webinar with AWS and TwelveLabs on driving fan engagement for live sports with multimodal and agentic AI.
• Databricks (technology)coreTechnology or IntegrationBeyond Databricks Ventures' investment, the partnership integrates TwelveLabs' video intelligence features into the Databricks Data Intelligence Platform.
• Snowflake (technology)coreTechnology or IntegrationBeyond Snowflake Ventures' investment, the partnership brings advanced video understanding to the Snowflake AI Data Cloud for Media customers.

Scale indicators15 records

Recent moves6 records

Expansion highlights6 records

Broad incumbents

• Google DeepMind (Gemini): DeepMind builds Gemini, a multimodal foundation model with native video understanding that TwelveLabs explicitly benchmarks against (Pegasus 1.5 vs. Gemini 3.1 Pro on Multimodal Prompting). Both target enterprise video understanding, search, and reasoning workflows, though Gemini is part of a much broader AI portfolio.
• OpenAI: OpenAI offers GPT-4V/Sora multimodal video understanding and generation capabilities that compete directly with TwelveLabs' Marengo and Pegasus models for enterprise video AI use cases. Both sell API-first multimodal video intelligence to enterprise and developer customers.
• Adobe (Firefly / Frame.io): Adobe's Firefly Video, Premiere Pro AI features, and Frame.io collaboration platform provide integrated AI video intelligence inside the dominant creative software suite. Adobe is the primary incumbent that Rodeo must displace in professional post-production workflows.
• Azure AI Video Indexer (Microsoft): Azure AI Video Indexer provides video understanding (speech, faces, sentiment, OCR) as a managed cloud service for enterprises, competing with TwelveLabs in regulated enterprise and media verticals. Both target large enterprises needing scalable video analytics.
• AWS Rekognition / Bedrock: AWS Rekognition and Bedrock-managed foundation models provide native video AI capabilities inside the AWS ecosystem where TwelveLabs is also distributed. As the host platform, AWS is both a critical distribution partner and a structural competitive threat.

Direct peers

• Runway: Runway builds AI video generation and editing tools for creators and studios, overlapping with TwelveLabs' Rodeo creative copilot and Pegasus video-to-text products. Both target media/entertainment customers with AI-powered video workflows.
• Synthesia: Synthesia provides an enterprise AI video generation platform with avatars and natural-language-driven video creation, overlapping with TwelveLabs' enterprise video intelligence and creative tooling use cases in media and corporate communications.
• HeyGen: HeyGen offers an AI video generation platform for enterprise content creation with avatar and multilingual video capabilities, competing in the same AI-video-for-business category as TwelveLabs, particularly around creative tooling and media workflows.

Emerging players

• Descript: Descript provides AI-powered video and podcast editing software with transcription and content-search features that partially overlap with TwelveLabs' multimodal video understanding and creative tooling (Rodeo) products. Both target creator and enterprise video workflows.

Others

• Mux: Mux provides developer-focused video infrastructure (encoding, streaming, analytics) that serves as an upstream component in many video pipelines that TwelveLabs targets. They are adjacent infrastructure peers serving the same developer and media customer base.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks6 records

Key highlights8 records

Customer concentration

• Xx.com/twelve_labs
• Facebookfacebook.com/TwelveLabs-61575755852747
• LinkedInlinkedin.com/company/twelvelabs

Revenue estimate

Valuation estimate

Number of profiles

Funding overview

Funding rounds10 records

Investors24 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Twelve Labs