TwelveLabs is a video intelligence API. It turns raw video into data that software can search, analyse and act on. Developers send video to the platform, which indexes the visual content, audio, speech and on-screen text together, so a team can search a library in plain language and find specific actions, scenes, dialogue or emotions without tagging footage by hand.
The platform is built around three models. Marengo is a multimodal embedding model that makes moments in a video findable by what happens in them. Pegasus is a video language model that reasons over the full timeline of a video, up to two hours long. Jockey is a video intelligence agent that answers questions across a library, and the page says it is available in Claude and ChatGPT.
The product is aimed at organisations that work with a lot of video, including media companies, sports teams, advertisers, public sector bodies and security teams. The page lists uses such as finding clips in archives, segmenting long-form video into scenes, scanning footage for compliance and brand safety, and assembling highlight reels from raw material.
Developers integrate through an API and SDK, and there is an MCP server for connecting AI tools to the platform. The company states that it is SOC 2 Type II certified and that data is encrypted.
Pricing starts with a free plan that includes 3 GB of storage and a basic workload allowance. The Plus plan costs $29 a month with 50 GB of storage, and the Pro plan costs $99 a month with 200 GB of storage and optional overage. Enterprise pricing is set by contract. Annual billing is shown at a 17% discount.
