Twelve Labs

Twelve Labs uses multimodal AI technology to let humans understand video in the same way they understand text, and to let humans search, generate text and classify video libraries. The technology is geared for businesses that have a lot of video and want to get useful information out of it, like media and entertainment companies, advertisers and surveillance companies.

Twelve Labs provides several APIs for different needs:

• Search: Search for specific moments in natural language, so you can find the exact moment you want in a large video library.
• Generate: Use prompts to create text about a video, which can be used for summarization, detailed reports, title generation, video highlights or chapters.
• Classify: Automatically label video content into predefined categories without the need for custom classifiers.

The APIs use state-of-the-art video foundation models that generate rich video embeddings that can be used for downstream tasks like search, generate and classify. The embeddings capture visual, conversational, text-in-video, audio and logo information.

Twelve Labs features include:

• Scalability: Can handle terabytes or petabytes of video data.
• Accuracy: Has been recognized by top researchers as one of the top performing AI models for video understanding.
• Customization: Models can be fine-tuned for specific content and domains.
• Security: Enterprise-grade security protects data and ensures compliance.

Twelve Labs pricing is based on the features you use:

• Marengo (Search & Classify): Includes free indexing for visual and conversation options, with other options costing $0.033 per minute or $0.0083 per minute for custom indexing. There are also infrastructure fees.
• Pegasus (Generate): Includes free indexing for visual and custom options, with prices starting at $0.001 per 1,000 tokens for input text and $0.002 per 1,000 tokens for output text.
• Compare Plans: Options include a free tier, developer plans with higher limits, including index limits, API calls and concurrent indexing tasks.

Twelve Labs supports several programming languages, including Python and JavaScript, so it can be used by a wide range of developers. The company is actively working on improving the technology with updates and open beta releases to keep it on the cutting edge of video understanding.

Published on June 14, 2024