Oxylabs Launches a Product Suite for Video Data
High bandwidth proxies, YouTube API & video datasets aim to satiate AI’s demand for multimodal data.
- Published:
Oxylabs, the Lithuanian provider of web scraping infrastructure and services, has launched a suite of products tailored specifically for extracting videos.
It includes three options: 1) high bandwidth proxies, 2) YouTube scraping API, and 3) pre-scraped YouTube datasets.
So far, all three have no public pricing and are sold through sales.
High Bandwidth Proxies
This is a proxy network built specifically for high-volume data collection. According to Oxylabs, its infrastructure can handle over 200 Gbps of bandwidth at once.
The network includes “millions” of stable IPs from diverse subnets. The provider doesn’t even mention which proxy types the pool consists of – the full focus is on capacity and scraping success.
High bandwidth proxies are designed to be a plug-and-play solution: they integrate using one dedicated endpoint, handling rotation and cooldown mechanisms automatically.
High bandwidth proxies are fully compatible with yt-dlp.
YouTube API
Oxylabs’ YouTube API returns videos and related data in a structured format. The product includes five endpoints:
- Search for discovery; it can fetch up to 700 results per query.
- Trainability to verify if a video is eligible for AI training purposes.
- Metadata to scrape the tags, supported formats, and other information related to the video.
- Downloader to get the actual video file – or files using batch downloads.
- Transcript to extract transcribed text from a video file.
The endpoints interplay with one another to cover all stages of YouTube data extraction:
YouTube Datasets
Oxylabs’ YouTube datasets cover over 4M videos from 1M channels, including their transcripts (JSON), metadata, video (mp4) and audio (m4a) files. All videos reportedly have consent for AI training.
The output can be delivered via webhook or to major cloud storage platforms. It’s also possible to request custom datasets.
Bottom Line
The video product suite has prominently appeared as a separate category on Oxylabs’ website. This shows how important – and data hungry – the use case of training multimodal AI models currently is for web data collection companies.
