Foundation Models
Data Infrastructure & Databases
TwelveLabs builds video-native foundation models that let organizations search, analyze, and generate structured data from large video libraries. Its Marengo model turns video into embeddings for natural-language search, and its Pegasus model turns video into text and timestamped metadata, both offered through APIs and on Amazon Bedrock. The company also sells applications built on those models, including the Jockey agent and a media compliance review product, for teams that would otherwise tag and review footage by hand.
We raised $100M to build Video Superintelligence
TwelveLabs raises $100M to bring superintelligence to AI video models
Pegasus 1.5: Deep Dive & Live Demo
A Guidance on Multi-Vector Video Search with TwelveLabs Marengo
Marengo 3.0: Real-World Multimodal Embedding AI
Nvidia-backed Twelve Labs is building AI that understands videos like humans