Twelve Labs
Category: AI Infrastructure
A multimodal AI infrastructure company building native video foundation models that enable developers and enterprises to search, classify, and generate text from video content with human-level understanding. Twelve Labs was founded in 2021. The company is led by Jae Lee. Based in San Francisco, CA, USA. Team size: 70-80. Total funding raised: $107M. Latest round: Strategic Round ($30M in Dec 2024). Key investors include NEA, NVIDIA, Intel Capital, Samsung Next, Index Ventures, Radical Ventures, Databricks, Snowflake Ventures, SK Telecom, HubSpot Ventures, In-Q-Tel (IQT).
- Founded
- 2021
- Headquarters
- San Francisco, CA, USA
- Team size
- 70-80
- Total funding
- $107M
Value proposition
Solves the 'video black box' problem by enabling machines to understand video context (actions, objects, sounds, and time) natively, rather than just analyzing static frames.
Products and solutions
Marengo 3.0 (Multimodal Video Embedding Model), Pegasus 1.5 (Generative Video-to-Text Model), Twelve Labs API & Playground, Mimir (Video Intelligence Platform)
Unique value
Utilizes temporal modeling to understand video as a continuous event with audio-visual alignment, rather than a sequence of static images.
Target customer
Developers, Media & Entertainment Enterprises, AdTech firms, Content Archives, and Security/Surveillance providers
Industries served
Media & Entertainment, Advertising & Marketing, Security & Surveillance, Education, Digital Asset Management
Technology advantage
Proprietary Foundation Models (VFMs) built from scratch for long-form video efficiency; Gated Modality Experts architecture.
How they differentiate
Native video understanding using temporal modeling versus traditional frame-by-frame image analysis.
Main competitors
Traditional Computer Vision Providers (Not explicitly named in source)
Key partnerships
NVIDIA, Oracle, AWS, Intel, Blackmagic Design, Vidispine, EMAM, Blackbird
Major milestones
Raised $50M Series A in June 2024, Raised $30M Strategic Round in Dec 2024, Integrated into AWS Bedrock, Launched Marengo 3.0 on AWS Bedrock at re:Invent Dec 2025, Launched Pegasus 1.5 in Apr 2026
Growth metrics
Team size grew to 70-80; Partnered with cloud giants (Oracle, NVIDIA, AWS); Named to CB Insights AI 100.
Market positioning
Specialized Multimodal Video Understanding Infrastructure
Geographic focus
Global (Headquarters in San Francisco, CA)
Patents and IP
Proprietary IP regarding Video Foundation Model architectures and multimodal embedding techniques.
About Jae Lee
Co-founder and CEO of Twelve Labs. Previously Lead Data Scientist (Sergeant) at the Cyber Operations Command for the Republic of Korea Ministry of National Defense. Software engineering experience at Amazon and Samsung Electronics.
Latest news about Twelve Labs
More AI Infrastructure companies
Official website: https://www.twelvelabs.io/