# TwelveLabs

> TwelveLabs is a multimodal AI company that builds video-native foundation models and provides a Video Intelligence API Platform. Developers use the TwelveLabs API to search, analyze, and embed video content using natural language and other modalities. The platform is built around two proprietary foundation models: Marengo (multimodal video embedding and semantic search) and Pegasus (video-language generation, understanding, and analysis). TwelveLabs enables developers to index video libraries at scale and query them with text, images, or composed queries — without manual tagging or annotation. Use cases span media and entertainment, sports, manufacturing, security, advertising, education, and enterprise video intelligence.

## Product

- [Product Overview](https://www.twelvelabs.io/product/product-overview): Describes the video intelligence platform architecture: foundation models, indexing pipeline, search and analyze APIs, and enterprise deployment options.
- [Video Search](https://www.twelvelabs.io/product/video-search): Documents the semantic search API: query video libraries using natural language across visual, audio, and spoken content with sub-second response times.
- [Analyze](https://www.twelvelabs.io/product/analyze): Covers the Pegasus-powered analysis API for summarization, chapter generation, highlight extraction, and open-ended video Q&A.
- [Embed](https://www.twelvelabs.io/product/embed): Explains the Marengo embedding API for building custom video applications: recommendation systems, content moderation, and similarity search.

## Models

- [Models Overview](https://www.twelvelabs.io/product/models-overview): Compares the two foundation models — Marengo (multimodal embeddings across vision, audio, speech, text) and Pegasus (video-language reasoning and generation).
- [Marengo](https://www.twelvelabs.io/marengo): Dedicated product page for the Marengo video embedding model. Processes visual, audio, dialogue, and motion across 36+ languages into 512-dimensional embeddings (6x smaller than competitors), handles up to 4 hours of continuous video, and indexes 30x faster than competitors while leading on accuracy. Includes composed image+text+audio queries, 5-sport recognition, and MCP server support for connecting Claude, Cursor, or other MCP clients directly to a video index. Achieves 73% composite performance vs. Google Vertex (52%) and Amazon Nova (55%), confirmed consistent with the Marengo 3.0 blog post below.
- [Pegasus](https://www.twelvelabs.io/pegasus): Dedicated product page for the Pegasus video-to-text model. Turns raw video into structured, timestamped JSON in a single API call with zero pre-indexing steps — handles up to 2 hours of continuous video across 12 languages, reads on-screen text and fine print alongside speech, and accepts a reference image (logo, face, product) as visual context. Also promotes MCP server support. Page-level comparison sets Pegasus 1.5 against Gemini 3.1 Pro and GPT-5.5 on max single-call duration and structured segmentation output, distinct from the Gemini-only benchmark comparison in the Pegasus 1.5 blog post below.

## Pricing & Enterprise

- [Pricing](https://www.twelvelabs.io/pricing): Lists Free, Developer, and Enterprise tiers with included hours, API quotas, model access, and support SLAs.
- [Pricing Calculator](https://www.twelvelabs.io/pricing-calculator): Interactive cost estimator based on video hours indexed, search query volume, and analysis API calls per month.
- [Enterprise](https://www.twelvelabs.io/enterprise): Enterprise offering — dedicated capacity, custom model fine-tuning, on-premise deployment, and 24/7 support.

## Solutions

- [Solutions Overview](https://www.twelvelabs.io/solutions): Index of industry-specific applications for the TwelveLabs video intelligence platform.
- [Media and Entertainment](https://www.twelvelabs.io/solutions/media-and-entertainment): Use cases for broadcasters and studios — archive monetization, content discovery, automated metadata, and editorial workflow acceleration.
- [Advertising](https://www.twelvelabs.io/solutions/advertising): Contextual ad placement, brand safety classification, and CTV campaign measurement using scene-level video understanding.
- [Government](https://www.twelvelabs.io/solutions/government): Investigative search, intelligence analysis, and public-sector video workflows with on-premise and air-gapped deployment options.
- [Security](https://www.twelvelabs.io/solutions/security): Surveillance review, incident detection, and security operations use cases powered by video search and analysis.
- [Sports and Broadcasting](https://www.twelvelabs.io/solutions/sports-and-broadcasting): Highlight generation, live and archived footage search, and broadcast workflow automation for sports content.

## Developers

- [Documentation](https://docs.twelvelabs.io): Quickstart guides, API reference, SDK documentation, concept explanations, and integration tutorials.
- [Developer Hub](https://www.twelvelabs.io/developer-hub): Entry point for API documentation, SDKs, code samples, and integration guides for the search, analyze, and embed APIs.
- [Sample Apps](https://www.twelvelabs.io/sample-apps): Reference implementations and starter projects demonstrating video search, analysis, and embedding workflows.
- [Research](https://www.twelvelabs.io/research): Published papers, model technical reports, and benchmark results for Marengo and Pegasus.
- [Whitepapers](https://www.twelvelabs.io/whitepapers): In-depth technical and business documents on multimodal video AI architecture and enterprise deployment patterns.

## Case Studies

- [Case Studies Overview](https://www.twelvelabs.io/case-studies): Enterprise video AI results across sports, media, commerce, and nonprofit organizations.
- [Qencode](https://www.twelvelabs.io/case-studies/qencode): Video encoding platform that embedded TwelveLabs directly into its transcoding pipeline, making every video searchable the moment it finishes encoding.
- [GS SHOP](https://www.twelvelabs.io/case-studies/gs-shop): Korean home shopping platform that layered TwelveLabs video understanding onto its recommendation engine, driving a 57.5% lift in ordering customers.
- [UNICEF Korea](https://www.twelvelabs.io/case-studies/unicef-korea): Nonprofit archive of over 8TB of field footage and images made fully searchable, cutting content retrieval time by 95%.
- [MLSE](https://www.twelvelabs.io/case-studies/mlse): Semantic video search applied to live and archived sports content.
- [SBS](https://www.twelvelabs.io/case-studies/sbs): Broadcast media video search and content management.
- [Dyn Sport](https://www.twelvelabs.io/case-studies/dyn-sport): Sports broadcast video intelligence.
- [Affiliate Network](https://www.twelvelabs.io/case-studies/affiliatenetwork): Video analysis for affiliate marketing and performance measurement.
- [Protege](https://www.twelvelabs.io/case-studies/protege): Video understanding for talent development and coaching.

## Partners

- [Partners Overview](https://www.twelvelabs.io/partners): Index of technology, channel, and integration partners across cloud, data, and media platforms.
- [NVIDIA Partnership](https://www.twelvelabs.io/partners/nvidia): NVIDIA strategic investor and infrastructure partner — covers NIM microservices integration and joint go-to-market.
- [AWS Partnership](https://www.twelvelabs.io/partners/aws): Available on AWS Marketplace and integrated with Amazon Bedrock for managed multimodal video AI deployment.
- [Databricks Partnership](https://www.twelvelabs.io/partners/databricks): Integration with Databricks Mosaic AI and Unity Catalog for enterprise video data pipelines.
- [Snowflake Partnership](https://www.twelvelabs.io/partners/snowflake): Integration with Snowflake Cortex for in-warehouse semantic video search, video-grounded RAG, and content recommendations without data egress. Snowflake Ventures is an investor.

## Resources

- [Blog](https://www.twelvelabs.io/blog): Technical blog covering model releases, research findings, customer stories, and engineering deep dives.
- [Introducing Pegasus 1.5](https://www.twelvelabs.io/blog/introducing-pegasus-1-5): Release announcement and benchmarks for the Pegasus 1.5 video-language model, including a 13.1% lead over Gemini 3.1 Pro on both segmentation quality (0.4279 vs 0.3370) and multimodal prompting quality (0.4555 vs 0.3243)
- [Marengo 3.0](https://www.twelvelabs.io/blog/marengo-3-0): Marengo 3.0 release notes covering the 73% composite performance benchmark (vs. Google Vertex 52%, Amazon Nova 55%) and updated embedding architecture.
- [Press](https://www.twelvelabs.io/press): Media coverage, press releases, funding announcements, and analyst recognition.
- [Webinars](https://www.twelvelabs.io/webinars): Recorded sessions with technology partners (Quickplay, Mimir, Adobe) and category education for media, archive, and advertising teams.

## Company

- [About Us](https://www.twelvelabs.io/about-us): Company background, mission, leadership, and advisory board.
- [Careers](https://www.twelvelabs.io/careers): Open roles across engineering, research, go-to-market, and operations.
- [Contact](https://www.twelvelabs.io/contact): Sales, partnership, and general inquiry contact channels.
- [Security](https://www.twelvelabs.io/security): Documents SOC 2 Type II certification, data handling, encryption, retention policies, and compliance posture
- [Patents](https://www.twelvelabs.io/legal/patents): Twelve Labs intellectual property.

## Key Facts

- Founded in 2021; headquartered at 55 Green Street, San Francisco, CA 94111 with R&D operations in Seoul, South Korea.
- Two production foundation models: Marengo 3.0 (multimodal video embeddings, 73% composite video-retrieval performance vs. Google Vertex 52% and Amazon Nova 55%; released November 30, 2025) and Pegasus 1.5 (video-language model, 13.1% lead over Gemini 3.1 Pro on both segmentation quality (0.4279 vs 0.3370) and multimodal prompting quality (0.4555 vs 0.3243); released June 1, 2026)
- CEO and co-founder: Jae Lee. Investors: NVentures (NVIDIA's venture capital arm), Databricks Ventures, Snowflake Ventures, SK Telecom, HubSpot Ventures, and IQT. CB Insights AI 100 honoree three consecutive years. Advisory board: Fei-Fei Li (Stanford), Silvio Savarese (Stanford), Jeffrey Katzenberg (former DreamWorks Animation CEO), Alex Wang (Scale AI CEO), Lukas Biewald (Weights & Biases CEO), Nicolas Dessaigne (Algolia founder), Jay Simons (Atlassian president).
- SOC 2 Type II certified; processes 10,000+ hours of video per day across customer workloads.
- Available natively on AWS (Marketplace + Bedrock), Snowflake (Cortex integration), and Databricks (Mosaic AI); deployable in shared cloud, dedicated cloud, private VPC, and on-premise configurations.
- Industry focus: media and entertainment, government, security, advertising, sports and broadcasting.

## Legal

- [Privacy Policy](https://www.twelvelabs.io/legal/privacy-policy): Data collection, processing, retention, and user rights under GDPR and CCPA.
- [Terms of Use](https://www.twelvelabs.io/legal/terms-of-use): Service terms, acceptable use, and customer obligations.

## Contact

- Website: https://www.twelvelabs.io
- Sales: https://www.twelvelabs.io/contact
- Developer Hub: https://www.twelvelabs.io/developer-hub
- Careers: https://www.twelvelabs.io/careers