Provider pricing

Twelve Labs AI API pricing

Twelve Labs builds video-native foundation models — Marengo (embeddings) and Pegasus (generative) — that understand motion, action, audio, and on-screen text natively rather than via frame sampling. The API supports semantic video search, classification, summarization, and chapter generation, priced per minute of video indexed plus per-token output.

Founded2021
HeadquartersSan Francisco, CA
Tracked modelsDirectory

Best for

Media, sports, security, and edtech teams building search or analytics over large video archives.

Strengths

  • Native video understanding — no frame-by-frame hacks
  • Marengo 2.7 embeddings for any-to-video and text-to-video search
  • Pegasus 1.2 generates summaries, chapters, and Q&A from raw video
  • Per-minute indexing pricing — predictable for large libraries